Observers cannot accurately estimate the speed of an approaching object in flight.

Objects approaching at the same speed, on the same trajectory, but at different distances from an observer, have different angular speeds at the eye. To recognize that the objects' approach speed is the same despite the differences in retinal motion, the observer must "factor out" the distance of each object. We examine whether observers can do so in three relative speed judgement experiments. In the first experiment we use a traditional psychophysical impoverished point-light display. In the second we use an un-typically rich cue-laden display. In the former case, observers are unable to accurately estimate speed, in the latter their performance is much improved. These two experiments, taken together, establish the range of possible performance. We then test performance in a display designed to provide the cues available in a typical natural ball-catching task. We find that observers are unable to make accurate judgements in this case. These results raise the question of how observers catch balls without accurate estimates of approach speed; we conclude with a discussion of potential solutions.