1. We have examined the responses of cells in the postero-medial and postero-lateral lateral suprasylvian areas (p.m.l.s./p.l.l.s.) in the medial and lateral banks of the middle suprasylvian sulcus of the anaesthetized, paralysed cat. 2. Visual responses were assessed qualitatively (for projected spot and bar stimuli) and quantitatively (for drifting, high-contrast gratings of optimum spatial and temporal frequencies, but varying in orientation and direction of drift). There was excellent agreement between qualitative and quantitative estimates of preferred direction of motion. 3. Comparison of responses to motion in the preferred direction and the opposite direction confirmed that the percentage of units with strong directional preference is higher in this region of cortex than in other cortical and subcortical visual structures so far investigated. 4. Cells were comparatively well 'tuned' for the direction of motion of a grating: on average the half-width at half-amplitude for the variation in response around the principal preferred direction was 23.2 deg for p.m.l.s. and 25.3 deg for p.l.l.s., thus falling within the range found for complex cells in area 17. In this and other aspects of their direction selectivity, neurones in p.m.l.s. and p.l.l.s. were very similar to each other. 5. Quantitative analysis of the direction-response functions revealed them frequently to be more complex than previously described. For many cells there were subsidiary response peaks in addition to the main peak at the principal preferred direction. Often there was an accessory peak 180 deg from the optimum direction (i.e. directional preference was incomplete), but there was also a clear tendency for responses to be specifically elevated for directions orthogonal to the principal direction. 6. Tests with stationary, contrast-modulated gratings of the optimum spatial frequency, but differing in orientation, revealed that neurones responsive to such stimuli (though limited to about half the population) were orientation selective, their preferred orientations for stationary and drifting gratings being very similar. 7. The functional architecture of the lateral suprasylvian cortex was studied by means of single and multiple penetrations at different angles to the cortical surface: in most penetrations in p.m.l.s. the preferred directions of cells tended to shift progressively in small steps across the cortex, at a maximum rate of about 360 deg mm-1. However, there were occasional 180 deg differences between neighbouring recording sites or between cells recorded at different depths within a single radial column.(ABSTRACT TRUNCATED AT 400 WORDS)