Whenever someone says what you have said, I know that there are really only two kinds of problems involved, if you have done as Will Robison has stated and have checked for debris in the carb and ensured that the fuel system is up to snuff.

An engine does not "know" the difference between bench running/ground running/flying in the air. Since it is not conscious and it is not aware of your intentions, there is no reason to assume that it is punishing you by not running right. You did not say such a thing, but I think that our emotions, as human beings, sometimes makes us feel this way, even though our intellectual/cognitive mind knows otherwise. So what's the point? Don't become frustrated and revert to your emotions for decision making. Let's think this thing through.

What changes when the model is in the air? For one thing, it unloads. Unloading means that the engine requires LESS fuel to run properly. As the fuel burns off, fuel draw drops, so leaving the ground slightly rich is a great idea, lest we end up running lean toward the end of the tank. So, the first possible cause is getting off too lean.

The second possible cause is fuel foaming. This one is as serious as a heart attack and is probably responsible for at least 75% of the posts that I see which refer to the engine quitting in flight.

The model sitting on the ground with the engine running and the model in the air with the engine running are two completely different creatures. Many times, in fact, the vast majority of times, the engine will run normally and the fuel will not foam with the model on the ground (or field stand). The fact that it is touching something solid changes the mechanical resonant frequency of the model, for lack of a better, engineering/more precise term. Once the model is "on the wing", this mechanical frequency changes. The latter is when fuel foaming usually occurs and throws the engine carb settings out of kilter.

One way to check for fuel foaming "on the wing" is to have the model running with a friend holding each wing tip and supporting the model up off of the ground. Go through the various rpm ranges and hesitate at the changed rpm for a while. Give the bubbles a chance to move up the fuel line and into the engine. Ah-ha! The engine leaned out or quit. What now?

You have to ensure that no part of the fuel tank, including the neck and brass fuel lines, are touching any solid part of the model. This includes the fuel tank neck passing through the firewall, or those funky firewall tank supports that are included in some ARFs. Yes, I do know better than the manufacturer in this instance. Their obligation is to their company, not to you. They do not care if your engine runs bad. After all, it was your crappy engine that crashed their wonderful model. Get it? <G>

The model's fuel tank needs to be supported by soft, uncompressed, foam rubber and must be completely prevented from touching any part of the model. This is a real PITA, isn't it? Sure is, but that is the way that it is.

I know, I know. You know so and so and his model has the fuel tank neck feeding through the firewall. He doesn't have any problems, so I must be wrong. Right? Wrong. He just got lucky. Even a stopped clock is right twice a day.