D. N. Lee (1976) described a braking strategy based on optical expansion in which the driver brakes so that the target's time-to-contact declines around a constant slope in the range -0.5 less than or equal to tau less than 0. The present results from a series of braking simulations confirm and extend earlier reports (E. H. Yilmaz & W. H. Warren, 1995) that performance is broadly compatible with the tau hypothesis. However, performance was not enhanced in situations that favored the estimation of tau, and unlike in earlier reports, performance deteriorated in the absence of a ground plane that provided information about speed and target distance. This finding suggests that the tau hypothesis does not provide a complete account of braking control.