On an elliptical treadmill a regular person can easily burn 1000 calories in one hour (treadmill reports calories burnt). This translates into: $$(1\times 10^3\mathrm{cal/hr}\times 4.2\times10^3\mathrm{J/cal})/3.6\times 10^3\mathrm{s/hr} \approx 1.2 \; \mathrm{kW} \approx 1.5 \; \mathrm{hp}$$ On the other hand, Wikipedia says "A trained cyclist can produce about 400 watts of mechanical power for an hour or more..." Is the problem that the treadmill gives wrong numbers? Or it is true that running using legs and arms - on elliptical machine or cross-country skiing which seems to be similar - a human can produce a lot more mechanical power than cycling? I thought the maximum power is set by the cardiovascular system, so it would be the same, running or cycling.

-
There may be a distinction between thermal power (what you've burned) and mechanical power (what can be converted to useful work). – dmckee Jun 2 '13 at 5:42
Yes, perhaps there is some kind of assumption that a calorie of mechanical work produced by a human requires several (3-5) times as many calories of food intake and/or fat burnt. So probably what Wikipedia quotes is the useful output energy (e.g. to power a human-powered aircraft) but what all those treadmills and sports magazines quote is the corresponding energy of food or stored fat. That would explain it. But what is the conversion factor for these calories? – Maxim Umansky Jun 2 '13 at 6:11
Also that Wikipedia quote seems a little suspect. They seem to claim that cycling is an order of magnitude more work than any other exercise. Is one of the world's most common forms of transportation really so superhuman an endeavor? – Chris White Jun 2 '13 at 18:34