Garmin Fenix 6 Pro Accuracy – GPS, GLONASS, oHR and Elevation

Reading Time: 12 minutes

Garmin fenix 6 specifications reviewHere are the results from my early ‘tests’ of GPS/oHR and elevation accuracy. I’ll endeavour to keep this post updated with new results for a month or so, so pop back for updates.

Return to the main Garmin Fenix 6 Review

FYI: Garmin fenix 6 specifications reviewWhen I got the watch, it immediately updated firmware to the production firmware ie v3 and GPS firmware to v2.5, so that might explain better results found by others in pre-production firmware and/or GLONASS/GALILEO might produce different results.

I always try to use GPS-only mode so that results can be compared between devices although, in Garmin’s case, it might also be fair to use GPS+GLONASS as Garmin seem to have tinkered with that over an extended period. I would assume that they are still tinkering with GLONASS and am pretty sure that they are REALLY tinkering with GPS+GALILEO ;-). There is little point in me testing stuff that Garmin may well change the next day.

Having said that my formal test did show GALILEO (71%, good) as being slightly better than GLONASS (69%, almost good) and GLONASS better than GPS-only (62%, average).

Let’s start off with an easy one

Garmin Fenix 6 Pro – oHR Accuracy – Swim (OWS + Pool)

This does NOT look too great and the Garmin oHR data is from the newly added oHR-when-swimming feature.

labelling error: it was GPS not GLONASS

The green line is based on my old HRM-TRI, which might be on its last legs. However, the HRM-TRI data does look right. I didn’t bother to line up the curves over time as they clearly wouldn’t match.

Here are two more that I would class as ‘acceptable’ and you could argue I should have had a third device. But I didn’t.

Here’s two more, the first in a pool shows the Fenix 6 +Elevate as being almost good but with the other one in OWS…I’ve no idea how to line these 3 up. The blue line is probably right.

This shows the danger of relying on ELEVATE when swimming. The tracks looks reasonable with no unusual troughs or peaks but it is 10bpm higher than HRM-SWIM – which I assume to be correct and which looks correct for this easy 30 minute swim

Garmin Fenix 6 Pro – oHR Accuracy – Bike

This are two fairly long rides and are good ones to use as there are some hard efforts in there as well as bumpy roads

Despite these tracks trending similarly to each other you can see quite big differences in the overall average HR. That may, or may not, be OK for you and will depend on the purpose to which you are putting the data. I’d remind you that the Firstbeat algorithms work a LOT on the HR data. (Garbage in…garbage out.)

Having said that, given the bumpy Surrey conditions, I’d say it was alright but that I’ve seen better…even from Garmin on very similar Surrey routes.

This was was segment hunting in Surrey so there are a few short and hard efforts in there and the Fenix 6 did well compared to the H10.

Next, we move to a TT bike with some fairly hard efforts on a reasonably good road surface, the Fenix should have done better here.

These two were two bike legs from the same brick session on a TT bike and both were at low intensity and each shared a similar route on the same kind of roads. As you can see one is inexplicably ‘not good’.

Here’s another that’s quite good until the end

Garmin Fenix 6 Pro – oHR Accuracy – Run

This was a fairly long and steady aerobic run. I’ve aligned the starting points of the curves to make it readable. The phrase “spot-on” comes to mind.

Here’s 2x 20 at sub-threshold running with some of it on grassland, again Spot On.

A similar 25+15 turned out equally good results, this one was a tad harder in hotter conditions.


This one is not great, however. I think I could have had the F6 tighter so, perhaps, as my arm got sweatier the slightly extra weight of the F6 combined with the sweat to give some movement and then it just became wrong. Incidentally, the extra 10 minutes of VO2max effort DID mess up my Firstbeat metrics. I also inadvertently paired the H10 to both the 935 and V800 so excluded the 935’s curve.

I need to be careful, I’m actually trying fairly hard at the moment as this one shows on a 4-3-3-1-1-1-3-3-4 run. Even here the F6 is not quite as good as the Forerunner 945 – even though the sensor is the same (presumably the exact same firmware commands as well). If anything I had the V800 banging onto the 945, so the 945 should have been worse. Maybe the weight of the F6Pro, whihc was on TIGHT, has something to do with the very sliht errors. NEvertheless, this performance is perfectly acceptable IMHO.

Finally, a nice long tempo run and the ELEVATE on the F6 mostly handles it well. Interestingly as it started to rain there were a few incorrect (orange) spikes from the F6.

Garmin Fenix 6 Pro – GNSS Accuracy – Bike

I don’t have any major issues with the GPS accuracy on the Fenix 6 when cycling. Sure there are a few feubles here and there but nothing I’m concerned about.

I’ll just give a few examples that support and contradict my lack of concern here 😉

This was a long ride and both the Wahoo and Fenix 6 were more than acceptable. Surprisingly I would say that the wrist-mounted Fenix bettered the ROAM on the handlebars. That should NOT happen Wahoo. Just sayin’.

The first of these two images shows the Fenix 6 beats the WAHOO ROAM (blue line) which cut corners, especially when there were trees involved. But that wasn’t always the case and there were occasions when the ROAM did better (second image)


These three images reflect a similar picture of the Fenix 6 (red) bettering the ROAM (Blue), as does the Forerunner 945 (green). A good cycling day for the Fenix 6 with GPS.

Another day, another ride but the same good results. One of the images shows the Fenix 6 seeming to go off the road but the deviation from the actual route is still well within an acceptable +/-5m. It jus tLOOKS worse when the track goes off the road rather than to the other side of the road. Fenix 6 (Green), Edge 820 (Red), Bolt (Blue)

This one was with GALILEO and the Fenix  (green) did very well for the entire ride. It was just this one small section (and another less serious one) in well over 60 mintues of riding that was thrown out by some trees. Such a small error is of little concern to me. (Edge in Red, Wahoo in Blue)

Garmin Fenix 6 Pro – GNSS Accuracy – Running

This is REALLY not great, with blatant erroneous tracks that anyone could spot.

Here we have red (Fenix 6), Blue (Ambit 3 Run – different day, best ever track), Green (Coros Vertix)


And again, #Pants

And again, although it’s strange that the Garmin and  Coros agree with each other. I was, for sure, exactly where the blue line is (+/-2m).

And again. It’s simply off-course at times like this

Despite the obviously numerous wayward wanderings, the Garmin did a good job at the overall distance getting accuracy to within 100.17%, although don’t expect to replicate that level of accuracy for a 10km run in a full built-up area.

The Fenix 6 also showed the instant running pace to be frequently >10 secs/km faster than it should have been with both GLONASS and GPS-only. Presumably, also 10secs/km slower in order to get the right overall distance but I only ever spotted it having me go faster…strange.

I might put some GALILEO charts up but, for now, the words are: GPS+Gallileo *DOES* seem to smooth the pace MUCH better than either GPS or GPS+GLONASS. Running with Galileo is sometiems actually GREAT but there are too many other times (in the same run) where the tracks are 5-20m off course.

Unfortunately for the Garmin Fenix 6, this was my ‘proper’ GPS running test and it scored 62% which is definitely POOR. In comparison, on a different day but exact same test methodology, the AMBIT got 85%. So a really quite big difference and this result is NOT up to Garmin’s normally ‘good’ standards. (Garmin never seem to score ‘excellent’).


Here is the methodology I follow, a results spreadsheet and all the source FIT files. You are more than welcome to come to different conclusions.

In fact, that run was not a one-off and I get similarly weird RUNNING results for GPS with the Fenix 6. It shouldn’t be as bad as it is and, as you’ll see further below it’s the same with GLONASS. Performance should be as good as the near-identical Forerunner 945…but it’s not.

Here we have the V800 (Blue), 945 (Green) and the Fenix 6 in Red. This first image is zoomed OUT a lot but you can still see the significant deviationS of the Fenix 6. There are some trees here but they’re not too challenging.

A little bit further on during the same run, there are some more dense trees and we can see again the Fenix 6 going off-piste when the other two agree. The trees here ARE more challenging and it’s difficult to know for sure the correct track (but it isn’t the red one!)

The rest of the same run included both challenging AND easy sections (first and second image respectively). and there’s no obvious winner but you can see the red line (F6) is more bouncy/jagged even when it is on the correct path.


Garmin Fenix 6 Pro – Running GNSS Accuracy – GLONASS

Perhaps the super-duper GLONASS is better.

Errr. Nope. Clearly not. The devices all were properly satellited up and the poor performance of GLONASS is towards the end of the run in any case. Although that did almost exactly coincide with going past Kiera Knightley’s parents’ old house (cnr Trowlock Ave). Maybe that was it? Fair enough.

V800 GPS Blue, 945 GPS Red, Fenix 6 GLONASS Green

A re-test over my standard route with GPS+GLONASS scored a modest improvement to 69%, which failed to beat the Coros Apex on the same run. Overall distance accuracy improved to 100.41% with GLONASS. (Both the 935 and 945 score about 75% with GPS-only, and I’d consider that performance level respectable).

On another day with the V800 and 945+GPS, the F6 Pro (Blue, GLONASS) is CONTINUOUSLY off the others. The 945 has the exact same chip. Other than the F6Pro’s metal casing I can’t see any reason for this persistent error as I get the same kinds of error with GPS-only as well. Elsewhere on this run the 4/5/6 year old V800 is clearly the best device…so much for old GPS-tech being rubbish.

I have some other runs but they don’t add anything to the above observations. If you want the most accurate GPS (or GPS+GLONASS) track then don’t buy a Garmin.

IMO, there’s something slightly wrong with running GNSS on this watch!

Garmin Fenix 6 Pro – Running GNSS Accuracy – GALILEO

Galileo isn’t notably different in my opinion from lots of runs HOWEVER the GALILEO test came in as the best and at a GOOD level.

In a tricky route next to house and, on the right, shops. You can see the 945 (red) really struggles whereas the V800 (blue) and Fenix 6 (green) are similar but the Fenix has me on the wrong (Northerly) side of the road.


The same colours/watches apply here on this narrow footpath under low trees. The v800 and Fenix 6 are very similar here.

But just before I got to that section there were more significant trees and the Fenix 6 did not do so well, as shwon here (green), where the V800 is clearly better.

I added the Casio WSD-F30 (green) into the mix and that has good GPS but here the Fenix 6 (red) almost keeps up with  it and I don’t know what the 945 (blue is doing) in htese relatively benign GNSS conditions

Further on in the same run the Casio is still doing very well and the F6 generally well, but struggling in places. Elsewhere (not shown) the Fenix 6 was OK but didn’t match the WSD-F30.

Here we have the Casio WSD-F30 (red), Fenix 6 (GLONASS, green) and Ambit 3 (blue). The tree lined paths were handled well by the Fenix 6 and Casio but the latter lost the plot when the signal was lost in an underpass on the 3rd image. Indeed these 3 iamges are fomr my GLONASS ‘test’ for the F6 and the whole track SUPERFICIALLY looks mostly alright but on closer inspection there were notable and persistent deviations throughout the run. Nevertheless the Garmin scored nicely enough.


GNSS Accuracy – Open Water Swimming

Garmin released some errant firmware earlier in the year which messed up the OWS performance. I’m happy to say that the Fenix 6 (red) put it a good performance here at Shepperton Lake and is very close to the normal/actual route taken. (Coros Vertix=Blue)

Here’s another one…pretty darn good.

Apex in Blue (not good), Fenix in Red

Next I turned Galileo on for the OWS and got a good track against the 945. However digging below the superfically good track it just looks like there are infrequently recorded location points – I’m really not that good a swimmer that I can swim THAT straight in OWS. The 945 (blue, GPS-only) doesn’t look quite right either but it is perhaps a better attempt at getting the times when I did deviate from a straight line inbetween the buoys.


Garmin Fenix 6 Pro – Elevation Accuracy

This looks good enough for me.

Some things to point out though.

  • The FIT file seemed to have no elevation in it at all. Strange. Yet there was elevation on Garmin Connect, so I exported a TCX from Garmin Connect and all was fine. I’m not sure what’s going on there (elevation correction disabled BTW). I’ve not delved deeper than that, maybe I made a mistake or there was some naming change in the file. Either way, the data looks cool
  • Inadvertently I used GPS+GLONASS on this one.
  • This was from a 5-hour cycle ride on a day with stable weather. That might explain why there was little drift in the elevation.
  • I’m not entirely sure if DEM data was used, I’m assuming not. The Garmin results do not tally with the ‘correct elevation’ track (from another DEM source) so I’m assuming the figures I show below are based on barometric altimetry-based, perhaps with a bit of GPS tinkering going on in the algorithm to periodically reset the elevation.


I’m a little disappointed here as the all round performances were not quite as good as I hoped.

Whilst the poor GPS & GPS+GLONASS running tests are final results I might do a re-test or two or try GALILEO. I sensed when looking at the watch that the oHR was better than it really was. Contrast that to the running GPS where I KNEW something was amiss from frequent consultation with the instant paces (testing a watch involves more than just sticking it on your wrist and looking at the data afterwards).

  • oHR performances seem to be going in the right direction. It still needs tinkering to beat the previous hardware iteration IMO
  • Elevation looks GREAT
  • GNSS/GPS looks mixed and in need of improvement especially with running

It’s still a super impressive watch though just buy STRYD and keep on using that chest strap if you want accuracy. 😉

Return to the main Garmin Fenix 6 Review






Support this site with purchases at these partners - should click to a local choice in your country

35 thoughts on “Garmin Fenix 6 Pro Accuracy – GPS, GLONASS, oHR and Elevation

  1. Interesting stuff.

    RE: Elevation missing

    There was a change in the MARQ time to how Garmin recorded elevation in .FIT files, specifically it’s now a 32-bit integer versus a 16-bit one. I only know this because it broke the DCR Analyzer and we had to implement a fix for it. I suspect that’s breaking legacy Sports Tracks.

    RE: Turning GLONASS off for testing

    I’ll be honest, I don’t entirely understand this choice. The default setting is GLONASS for mostly good reason: It’s better. Garmin has spent all their effort on optimizing this (not Galileo, nor base GPS mode). And when it comes to comparing against others, everyone else is using that too.

    RE: Loading maps

    Where are you getting that? Are you talking just when turning on the watch, or? That’s the only time I ever see that in normal use. If I’m loading the map data field its instant, and opening up a course with maps takes about 0.5 to 1 second in total. Just trying to understand where you’re seeing that.


    1. elevation/ST3.1 … yes that sounds right. which is very annoying
      gps/glonass – it’s better when it’s better and not when it’s not. the glonass mode incorporates all the gps goodness, which, as i showed, isn’t good. (anyway, let’s not go there)
      loading maps – IIRC I have several map regions loaded on the device. the ‘loading maps’ happens on the 945 as well but it’s worse on the fenix. i may have less maps on the 945. i’ll pin down the scenarios

      1. I know it seems to be a true battle on the internet as to what GPS setting to use, but I used just GPS mode for a downtown run (albiet not in a city that has a ton of tall buildings) as well as an open water swim last night. The tracks were pretty stinking good. Only when I was running on the sidewalk against a 5 story building did the GPS track ‘jump’ to show me running through the building a few feet. Other than that, Im pretty impressed by its accuracy. The OWS track was miles and miles better than my 5+.

        I’ve tried using GPS+glonass and GPS+galileo (neither of which should help me in Oklahoma, USA), and those settings always seems to make things worse.

        I did notice that when I uploaded my run to Strava, It does not allow me the option to correct elevation. I used to do this all the time on my 5+, especially when running hill repeats on a local bridge. Not sure if this is a strava or garmin thing…

        1. strange. you find the same as me?
          glonass and galileo satellites hould be over you? i think the galileo ones should not work for you too
          strava, i believe, corrects elevation to what it thinks it should be for each gps point. you can enable elevation correction in garmin but that is different (tho the end result should be the same or similar)

        2. Both Glonass and Galileo are global satellite systems and have full coverage over Oklahoma.

          > I did notice that when I uploaded my run to Strava, It does not allow me the option to correct elevation.

          This is interesting. Strava “trusts” devices with a barometric altimeter and offers their correction only to those devices. For devices w/o a barometer Strava automatically “fixes” the elevations (and thus doesn’t give you the option). I wonder if Strava is not realizing what your Fenix6 is yet… An easy test is to look at the elevation plot on an out an back activity – if using Strava’s elevation model it will be Perfectly symmetrical, if using the Fenix6 barometer it will be close but have deviations and/or drift.

    2. I lost sleep about the GLONASS last night (thanks Ray ! 😉 )

      So I did a +GLONASS recovery run today and included the results above.

      GLONASS is worse on this occasion. I have many previous occasions.

      Interestingly I was running at 5:00/km but the Fenix 6 had me going at 6:00/km for VERY extended perios, sometimes slower. Units are set to KM. I must have a bad unit.

  2. That’s a rough score. Seems no one has been able to quite figure out the Sony chip. To be fair, a bit of a moving target as they’re also releasing constant updates for it.

    1. quite so.
      well…if they are releasing updates then it can’t be ‘there yet’. GPS is likely to be the most stable and hence the most useful for me to use for comparisons. there’s no point in me doing one of these ‘formal’ tests and spending 3 hours in total only for the firmware to be changed the next day. I have a real job and a life as well as this.

        1. hmm. today’s test was also for the Vertix…i think it was slightly worse. 🙁 I have had better results previously, today could have been a bad satellite day, i’ve not added those, stats VDOP etc, to the spreadsheet yet

          1. Well…dang lol.

            If Suunto decides to enable route planning through their new app for Ambits it may well get dusted off and brought back out

      1. C’mon, man! I’m still looking for reasons to buy one of these and you’re not helping!

        Seriously though, I sort of expected this. It would have been amazing to me if Garmin was able to produce more reliable GNSS tracks with a new(ish) chip in a new watch. I’m not at all convinced these Sony chips are going to be at all reliable. I guess time will tell.

        On the other hand, I had an awful track today with my Fenix 5. Perhaps I should lower my standards and be happy with what I have.

        1. i did have a 945 at the same time as the F6 and the 945 was better. Also 945 was better last week in some other workouts i’ve not shown (yet)
          I’m NOT saying this is anything to do with the case of the f6 or anything like that. garmin WILL improve glonass and galileo over time, as dcr implies

  3. Please, TFK, some comments about the weight of the Fenix6 while running in comparison with the FR945. Do you find that the watch stays in place well or bounces a bit more than the FR945 (that is perfect for size and weight).
    I really like my FR945, but the wider screen of the Fenix6 makes me evaluate if it could be worth a replacement (if I eventually manage to sell the FR945). It’s a pity that the wider screen of the Fenix6 didn’t appear in the previous watches FR945 and Marq.

  4. To help DCR understand the “Loading Map” issue: Map/Around Me/Loading Map screen up to 30 seconds.

    This happens whenever you try looking around your area.

  5. FWIW, my 6s has so far been the most accurate Garmin I’ve ever owned (and I’ve had most of them since 2013). It’s on par with the TomTom Runners, which was an extremely accurate watch.

    1. TomTom’s were good. I did an interesting soak test once.
      placed lots of devices (recording) in the middle of my garden. As expected all the recorded tracks were all over the place…front garden, neighbour’s garden. EXCEPT the tomtom which drew a stright line about 4m long (or something similar). I live near one of the ex-TomTom engineering guys and we had a beer once, he explained how awesome they were.

      1. Such a shame. Wish everyone in sports tech started from the premise that the basics have to be flawless before you start adding features.

        1. such a customer-focussed approach seems always destined to fail ! 😉

          when I last looekd it was something like 2019 and I can’t think of more than about 2 running watches that can accurately tell you how fast you are runnign at right now. And neither of them are as accurate as STRYD.

          oh well.

      1. I have an XC circle I’ve been riding 100s times with majority of it exacly (+/- 20cm on the same path). with my v800 on my wirst and wahoo on the bar i am seeing 2 things:
        1. v800 is far more consistent over the same training (e.g. 3hrs riding the circle x6 – x10 times) the track looks closer lap after lap
        2. distance per lap is consistent (+/- 30-40m on 4.58km) over time (e.g. it doesn’t really matter if this is autumn or middle of vegetation season)

        Only one thing which is consistent between these 2 units is ascent and descent.

        I’d like to mention that the path registered with my Samsung S8 (super duper using all sat systems) installed on the bar is FAR from any of these 2 units.

        After all i am struggling to understand why any of these devices “seeing” more triangulation points than v800 is worse in the same conditions?

        Just to compare – when i’ve been ridding on a road in rural areas i can spot the side of road i was on in case of both wahoo and v800 – almost no difference there.

        1. +/- 20cm is extreme accuracy.
          GPS is accurate to about +/-5m.
          Yes the V800 can be awesome (I’m using mine again at the moment….memories!!)
          trees can make some difference…much less than buildings
          are you sure you have all the correct settings on the S8? IDK that phone but some phones are awesome and some are not. maybe a-gps was not laoded and you got no initial fix…
          read this:

          1. I think the poster said their are biking on the trail within +/- 20cm.

            That said, as I just posted, Garmin told me that for consumer GPS devices their standard is 15m accuracy, not 5m. So tracks can deviate up to 30m from each other (100ft!!!!) and Garmin would consider it to be a-OK.

            Not OK with me.

  6. My observations are similar to yours. The GPS track is absolutely horrendous, probably the worst I have had on any device I kept (starting the Motorola MotoACTV!), yet, surprisingly the total distance is about right.

    Case in point, I was running a marathon, mostly open/forest area in Maine, easy reception. Or easy for everyone else. I picked a dozen or so other runners around me in Strava flyby. Check it out, the Garmin Fenix 6 Pro is the black line

    During the entire 26 miles it is almost always the outlier. It is *BY FAR THE WORST* GPS track of all of them.

    Let’s face it, there are far better smartwatches out there and the screen technology is just not what one expects in 2019. The raison-d’etre, the primary reason for it’s existence, is to be a sports GPS tracker. Yet, this brand new, premium priced device from the market leader MISERABLY FAILS in that regard. It’s sad and I feel like a fool for purchasing it.

    1. Let me follow up on my own post with the resolution of this. I called Garmin support and I got someone who went back to engineering and showed them my GPS tracks. Here is their official statement:

      This is a consumer device and Garmin considers consumer device GPS deviations of 15m to one side or another as within specs. In other words, on runs back-and-forth, the tracks can deviate up to 30m from each other to stay in Garmin spec. They looked at it and said that my Fenix 6 Pro was just within that.

      So here we have it. Garmin’s premium device has an accuracy of 15m.

      There also was a strong suggestion going with a Garmin 945 instead for running activities as the plastic case will yield better results in comparison with any device with a metal bezel.

      Personally, I do not consider 15m GPS accuracy as a reasonable spec for premium devices in 2019. What 15m inaccuracy will do to your pace calc, is anyone’s guess. That said I am grateful for Garmin support coming clean on this.

      1. blimey, i’d like to see a transcript of that chat on th etwo points you raise!
        it back up what i say about PACE, you need a footpod if it’s important to your

        satellite accuracy is +/-5m ie that’s the best possible, so i guess what Garmin are saying about +/-15m could be true

Leave a Reply

Your email address will not be published. Required fields are marked *