We have measured orientation-discrimination thresholds of 4 deg in the cat, confirming an earlier study of Vandenbussche and Orban (1983). Unlike Vandenbussche and Orban (1983), we found that orientation-discrimination performance is not better at principal, as compared to oblique, reference orientations (no oblique effect). Despite the absence of the oblique effect, and despite the discrimination thresholds which were elevated by a factor of 4 compared to humans, orientation-discrimination performance of cats and humans is qualitatively similar in a number of aspects. First, orientation-discrimination performance as a function of length and contrast is qualitatively similar to human performance. Second, as in humans, detection and discrimination of the stimuli are closely related. Finally, randomizing the contrast between the stimuli does not affect orientation-discrimination performance. This suggests that similar computations underlay orientation-discrimination performance in both species. In summary, our results confirm that the cat is a useful model for human orientation-discrimination performance.