How accurate is the 'calories burned' figure on my GPS device?

I use the Garmin Edge 200 to record the route, distance and speed of my rides. It also gives me a prediction of the amount of calories I have burned on any particular ride.

The only data I have given it is my weight. It must use the speed and elevation GPS data to calculate a prediction but id like to know how accurate that can be if it doesn't take things like wind strength & direction, quality of surface and tyre type into consideration.

As an example, I went for a 50km ride today and it informed me I burnt 1200 calories. Is this in the correct order of magnitude?

• It might be correct within a factor of two, in not-too-extreme cases. May 2, 2015 at 14:19
• Did that 1200 include your Basal Metabolic Rate (What you would have burned sitting on the couch). Toys that provide Calorie burned provide "an interesting number that quantifies, without evidence, what you want to know.". i.e. Ignore it for scientific purposes, quote it for bragging rights. May 2, 2015 at 20:37
• General rule of thumb is 100 calories of food per hour for rides over an hour or two long. For short rides under an hour, no food required unless you're going really hard. I did 72 km in 2800 kJ with 1690 m of vertical, so 1200 kJ on the flat over 50 km sounds about right. Try strava app on a fancy phone, for a comparison.
– Criggie
Jan 12, 2016 at 7:21

Garmin has offered several different algorithms over the years to estimate energy expenditure and some are better than others. However, the only way to get really reliable calorie burn data from a bike computer is with a direct-force power meter, and -- in my experience, at least -- most simple algorithms to estimate calorie burn will overestimate the calories you've expended by 50% or more.

Unlike many other endurance sports (running, swimming, cross-country skiing), most people's efficiency while cycling falls into an extremely narrow range. I've seen elsewhere that the range is typically between 20-25%, meaning that 75-80% of the energy you expend while cycling is not applied to the pedals and the remainder is. Since there are 4 kJ per dietary calorie, the metabolic cost to you of applying 500 kJ of energy to your bike's drivetrain (i.e., averaging just under 280 W for 30 minutes) is thus likely to be somewhere between 500 and 625 kCal.

Unfortunately, the Edge 200 won't record power data even if you have a power meter. If you want a better approximation of your energy expenditure, you might consider looking at something like Strava's estimated power instead. This estimate still doesn't account for drafting and winds, but it uses grade, acceleration, your weight, and the weight of your equipment and is thus more accurate than estimates based on speed and athlete weight alone.

• I had noticed the rather significant discrepancy between the Garmin calorie value and Strava's prediction. For the same ride (uploaded directly to strava) it predicts just over 700 calories. Closer to what I would expect, but still rather high! May 6, 2015 at 18:22
• Strava is dreadful at elevation changes. This track was at sea level and no more than 50 cm variation. However it shows a maximum elevation of 62 metres ? strava.com/segments/11048630 Still its free, and theres nothing wrong with a backup.
– Criggie
Jan 12, 2016 at 7:16
• Yes, GPS-based elevation can be a lot less accurate than a barometric altimeter; it looks like you recorded that with a phone app instead of a cycle computer with an altimeter. Most methods of recording elevation over a ride have critical flaws, though (barometric altimeters are probably the best option but they are sensitive to mid-ride weather changes, etc). Jan 13, 2016 at 18:17

By order of magnitude if you mean by a factor of 10 then most likely yes.

You have wind, rolling, and gradient resistance.

It is going to be spot on for gradient resistance so if you have a hilly ride it will be more accurate.

Clearly it does not account for wind.

I expect it assumes a mid level road bike on decent pavement.

At speeds over 10 mph rolling resistance is the smallest of the three.
So it is going to be off on what is typically the smallest factor.
See the graph in the link:
http://www.schwalbetires.com/tech_info/rolling_resistance

On a hilly road ride with no wind it is probably with 10% overall.

Riding flats drafting in a group it is going to be off by a lot.
Trail riding it is going to be off by a lot.

I'm using a Garmin 510 along with HR strap and Stages power meter on my road bike, and even with all that data I think it still over-estimates calories burned by nearly double. I arrived at this figure based on tracking my body weight and calorie intake over many rides and months, taking into account hydration etc as well. I'm 72kg, 9% body fat and a CAT2 racer doing 10+hrs/wk, and eat very cleanly. Strava seems to derive an even higher calorie burn (~10%) than Garmin, but not consistently higher. Overall I dont think these calorie burn numbers are worth using to manage energy intake or your weight. I have no doubt that the accuracy varies hugely between individuals and circumstances due to variables such as body composition, metabolism, riding efficiency (a big one), temperature, sweat rates, diet and a whole bunch of other factors. About the only value I see in calorie burn numbers from devices is bragging rights over coffee at the end of a big group ride:-)

• Gidday and welcome to SE Bicycles. Good first answer, plenty of relevant details without getting distracted. I look forward to your future contributions.
– Criggie
Jan 12, 2016 at 6:40
• As far as I can tell, Strava calculates calorie burn for activities with a power meter by adding approximately 10% to the kilojoules expended over the ride (presumably assuming ~22.5% efficiency). But calorie counts for foods are typically not particularly accurate, and even if you're weighing your food you might be getting more or less energy than you assume. (And factors besides energy balance also impact body weight!) By the way, the sport science literature disagrees that cycling efficiency accounts for a big difference between individuals: d3epuodzu3wuis.cloudfront.net/R060.pdf Jan 13, 2016 at 18:14

I have a Garmin 800 which includes heart rate in addition to speed and elevation. Heart rate tells it what my actual level of effort is, so that should take wind and all those other variables into account. It also knows my age, gender and weight, so with all that data it should be able to estimate calories burned pretty closely, or at least as closely as is possible outside a laboratory.

Well, according to my Garmin I should lose almost 20 pounds every year during riding season. Sadly, I must report that hasn't happened yet. Not even close. Yes, my calorie intake and levels of other activities during riding season remain the same, so if I'm burning, say, 3500 calories per week riding my bike like Garmin says, then I should be losing a pound a week, give or take*. In practice, though, I find that I lose maybe 5 pounds over the entire season, which is 75% less than Garmin predicted.

• I say "give or take" because weight loss and calories aren't a simple equation. There are dozens of variables involved and no two people are the same, but the margin of error should be somewhere in the 10-20% range, or at least less than 50%, and definitely not the 75% error I see consistently.

I therefore conclude that Garmin and the other half-dozen fitness apps/gizmos I've tried all grossly exaggerate calorie usage. Why? Because looking at your expensive gizmo and seeing a huge number of calories burned makes you happy and sells more product.

So, based on my admittedly casual and anecdotal observations, I would recommend subtracting 50-75% from whatever Garmin says for a reasonably accurate estimate of calorie burn.

• Does it definitely use heart rate in the calories calculation? I looked into this a while ago and the equations at the very least need your resting heart rate - if it doesn't take your resting heart rate into account I would suggest that the heart rate output is independent of the calories which are calculated from gps. May 3, 2015 at 7:54
• On another note I would expect them to use a simple algorithm (probably over-simplified) but pretty realistic to slightly optimistic. It may be assuming and average bike to account for road and mountain bikes, in which case you'd be more efficient. When I tried calories in vs. out using a range of measuring exercise equipment it was certainly effective but estimating the intake was much harder, and prone to errors. May 3, 2015 at 7:58
• To be honest, I was expecting it to be an over estimate. I just want to know how many calories I need to consume to make up for those lost on long rides. Or at least to make sure I don't consume too many! May 3, 2015 at 10:54
• Remember, muscle is denser than fat, so you can loose fat and gain muscle mass, while holding the same weight and reducing your diameter at the same time.
– Criggie
Jan 12, 2016 at 7:17
• @Criggie I understand that and go by more than just weight. I pay attention to how clothes fit and also body fat % as measured by calipers. Jan 12, 2016 at 15:09

In my opinion it is only a usable value data point to compare against the same previous data point. An unfit person will burn more calories at a given level of exercise (based on heart rate and other factors) than a fit person. An unfit person will have a less efficient heart and circulatory system. A calorie usage point can't be made without calculating the level of exertion.