Four experiments intended to show instrumental escape learning in the housefly (Musca domestica) were performed. The general procedure of each study allowed subjects to move from a hot, brightly illuminated half of a Plexiglas tube to the other half, which was relatively cool and dark. In Experiment 1, subjects were able to move freely throughout the tube at all times. The amount of time spent in the lighted half decreased significantly over the course of twenty 1-min trials. In Experiments 2, 3, and 4, a gate separating the two halves of the tube was used to create a typical shuttle box. In Experiment 2, escape latency decreased significantly over trials, and the effect was maintained over a 24-hr intersession interval. However, the effect was reversed when heat-absorbing filters were inserted. Experiments 3 and 4 showed that repeated exposure to either heat or light was not sufficient to produce the effect. Escape latency decreased only when there was a contingency between the response and escape from heat.