Single striate cortical neurones were recorded from adult cats, lightly anaesthetized with N2O/O2/halothane. The receptive fields for the dominant eye were subjected to direction-specific adaptation by a square-wave grating of optimal spatial frequency and velocity, drifting continuously in each neurone's preferred direction. Recovery of the neural motion after-effect induced by prior adaptation was assessed with the same grating pattern which now moved alternately in the preferred and opposite directions. In controls the same tests for recovery followed a period of exposure to a uniform field of identical luminance to the adapting grating. Three sets of measurements were made to establish whether the adaptation was orientation- as well as direction-specific. In the first, test grating orientation was maintained constant and optimal for each neurone whilst adapting orientation was systematically varied. In the second, test orientation was varied whilst maintaining adapting orientation constant. In the third set, adapting and test orientations were initially fixed at each neurone's optimum; they were next set, non-optimally to one side of the optimum. Results from the latter configuration were compared with similar tests in which the test grating remained at that non-optimal orientation whilst the orientation of the adapting grating was now altered to a new point on the other flank of each neurone's orientation tuning curve that was matched for strength of adaptation. Thus the degree of adaptation was identical in each case, but zero orientation difference between adapting and test gratings in one case was contrasted with a substantial orientation difference in the other. The results from all three sets of data were unequivocal: in simple neurones, and in standard and intermediate classes of complex neurones, but not in special complex neurones, the sequential effects of adapting gratings on the responses and sensitivity to subsequently presented test gratings were maximal when their orientations were matched and optimal for each neurone, less marked when orientations were matched but non-optimal. In conclusion, adaptation induced by pattern motion was orientation- as well as direction-specific only in standard (length summating) and intermediate complex neurones, and in simple cells; in special complex neurones it was not.