During locomotion, observers see a characteristic pattern of motion referred to as an optic flow field. To investigate how they make use of this pattern, we have developed a paradigm for testing visual function during locomotion. Foot placement was recorded while cats walked down an alley cluttered with a high density of small objects; the task was to avoid stepping on any object. In the experiments reported here, motion cues were eliminated by the use of low-frequency strobe lighting. In bright continuous light cats performed with great accuracy, and likewise at scotopic light levels. However, in strobe lighting their error rates increased more than threefold. This deterioration could not be attributed to lower acuity, since the cats' performance remained excellent when the light level was reduced well below that afforded by the strobe light. When very dim continuous light was combined with low-frequency strobe lighting, performance was substantially better than under strobe light alone. We conclude that motion-sensitive neurons make a major contribution to visual guidance of foot placement during locomotion. When strobe lighting is combined with very dim continuous light, even the minimal motion information available in the intervals between bright strobe flashes improves performance significantly. Cats were also trained to discriminate between complex patterns, and this discrimination was not affected by strobe lighting, suggesting that motion-sensitive neurons are not critical for this analysis.