In calculating the precise speed of an object, the visual system must integrate motion measurements across time and space while keeping motion measurements from different objects separate. We examined whether an initial coarse estimate of local speed may be used to segregate the motions of different objects prior to a precise calculation of object speed. Our stimuli consisted of 256 dots that moved upward at two speeds. In Expt 1, each dot alternated between the two speeds every 133 msec. When the speed alternations were asynchronous across dots, subjects saw two transparent surfaces moving at different speeds and their ability to discriminate changes in the slow speed were unaffected by the presence of the fast speed. This experiment suggests that before integration, motion measurements may be segregated according to speed. We sought more conclusive evidence for this claim in Expts 2 and 3. In Expt 2, dots with 33 msec lifetimes were used to generate the two speeds. Although individual dots permitted only crude speed discrimination, subjects perceived this stimulus as two surfaces moving at different speeds and they precisely judged the slower speed. Apparently, the coarse local signals generated by the slow dots were segregated from those of the fast dots and then separately integrated to produce a precise speed signal. In Expt 3, the dots again moved at two speeds, but each speed was generated by a range of spatial and temporal displacements. Once more, subjects saw two surfaces and precisely judged the speed of the slower surface, demonstrating that segregation may be based solely on differences in local speed. We conclude that the visual system calculates two speed signals, one speed signal is coarse, local and used for segregation and the second signal is precise, global and used for speed discrimination.