Maybe I can help explain it.

On a treadmill, you are actually running a certain distance as displayed on the treadmill computer unit. Crank the incline up to 2-3% and you're not just skipping up and down, but actually propelling yourself up and forward and doing the actual work.

With an exercise bike, you don't know what the gear ratios are. As it is, there is only one gear on an exercise bike. When you pedal, and then select increase or decrease resistance, the computer raises or lowers the resistance being applied to the flywheel. However, after time, the unit will need to be recalibrated, the resistance unit replaced, etc. With most exercise bikes, this is not done in a timely manner, if at all, so you have worn out parts and uncalibrated computers trying to tell you how much work you're doing, when they aren't accurate themselves. Thus, even though you might be going very hard, the data output from the computer head is not at all realistic (it could be too high or too low) in most cases.

Treadmill miles usually count as miles but they are biomechanically very different. The 2% incline is used to counteract the effect created by the belt propelling itself forward, but it is an imperfect solution. If you look at the limited amount of academic research comparing outdoor/treadmill running, you'll see that the consensus is basically that they are similar, but not identical, activities. You are covering miles, but you're covering them in a biomechanically different way than you would outdoor running. It would be akin to comparing running and walking -- miles are miles, but the physical load is profoundly different, particularly over long distances.

In cycling, all metrics come with qualifiers: when you talk about mileage, one needs to know about the terrain, wind conditions, etc. that came with it, because they make a huge difference on the level of effort involved. No one talks about average speed (except n00bs) because the number of additional considerations render the number meaningless (more or less). You can hum along at 20mph outside for an hour, and single two-minute traffic light will take two-thirds a mph off your average. It gets "worse" if you start factoring in real conditions -- but it isn't "worse" unless you are trying to compare your average time against an impossible-to-achieve standard.

Bottom line is: No outdoor cycling course will ever come close to being like sitting on a stationary bike in a gym, so why ask how your ride on a stationary bike in a gym rates against riding in real conditions outdoors?

i just wrote a long reply, realized I was wrong and now contemplating some masochistic act or other. well then. sh*t

First of all, wind resistance is a HUGE factor in road riding, and as your speed increases so does its importance. When you run the speeds are not nearly as high and the aerodynamic difference between running on a treadmill and in the real world are not as big.

I think what you do not seem to grasp is that there are three different arguments in play here. These really have to be separated out to have a reasonable conversation.

Also for the purposes of this discussion, "indoor riding" is anything where you aren't actually moving forward. A gym bike is one of those big gaudy things with little adjustability and lots of instrumentation, and a spin bike is similar but is usually much more adjustable and has no instrumentation, and is usually used in a group (class) environment. Trainers use a real bike and provide support and resistance so that you can ride your bike without going anywhere. Usually the resistance adjustment is very rudimentary and there is no feedback or instrumentation. Computrainers actually connect to a computer and can simulate real riding more closely, as well as provide instumentation, including your power output and an approximation of your speed.

1) Is "indoor" riding worthwhile for training?

Yes indoor riding is perfectly worthwhile training. You can get a good aerobic workout on gym bikes or spin bikes or trainers. Spin classes in my opinion are also pretty good for anerobic workouts (intervals) because the structure of the class help to push you. You may have the discipline to do this on your own on a gym bike or trainer; I do not. A computrainer wout also probably help with this.

2) Can you count gym bike or trainer distance for training purposes?

I would say that a gym bike's distance should absolutely not be counted. THe numbers that they produce are a complete fantasy that are designed to make people feel good about exercising. As for whether you count trainer distance, that is up to you. Many people do, but I don't; it just counts as exercise time.

3) Can you compare gym bike or trainer performance to anything else?

No. At best you can compare a gym bike to that same exact bike. Not even to the same model bike sitting right next to it. A trainer you can compare power readings. You can probably compare Computrainer power readings to on-bike powermeters such as Powertap or SRM, but beware that there is often variation even between these different types of units. You most certainly cannot compare SPEED measured on a gym bike or trainer to any real-world riding. A Computrainer may give you a better approximation because that is what it was designed to do.

But you cannot compare your performance, even riding a computrainer, to actually going out and riding the course for real. period. end of discussion.

Everybody is talking about a resistance trainer. A fluid trainer or a wind trainer, if the resistence is designed in correctly (i.e. the drum and paddles inside the resistance unit are the right size), can be pretty good at simulating a real road ride on flat ground. Because it is still fluid resistance, the resistance will vary with the square of speed, just like on the road, giving the cubed power law.

I have a blackburn fluid trainer (BTW, buy something different, the attachment mechanism on this particular trainer kinda sucks) and on a roadbike in the 53/19, my speedometer has me going at about 30kph on a constant power basis. I can maintain a slightly greater speed on the road (on flat ground and without a head or tail wind - maybe 35kph with the same effort), but it is fairly representative of road riding. When I shift up for an interval, the power vs. speed output is pretty realistic.

The only difference is that the acceleration and deceleration of the wheel is different. If you accelerate on the road, you are fighting the momentum of the wheels and the mass of the bicycle as a whole. If you accelerate on the trainer, you are only accelerating the one wheel, which means it is easier to accelerate on the trainer. On the other hand, because energy is being stored by the momentum in one wheel only, if you have a choppy pedal stroke, the trainer will magnify this because the wheel want to decelerate much faster than the entire bike would out on the road.