Three experiments were performed in which a stimulus with homogeneous color and luminance was masked by a second stimulus containing contours. In the first experiment the target was a large white disk and the mask was a white circle concentric with the disk but of smaller radius. We found that the mask had a large (up to 2 log unit) suppressive effect on the brightness of the target, but only inside the radius of the mask. With monoptic presentation of target and mask, the greatest suppression was observed with an SOA of 50-100 msec. With dichoptic presentation the strongest suppression was obtained with simultaneous stimuli. The second experiment demonstrated that the latest time at which masking was effective was correlated with the distance between the edge of the target stimulus and the contour in the mask. One possible explanation of the results from these two experiments is that the masking contour is interfering with the propagation of a brightness signal from the target's border. In the third experiment gaps were introduced into the masking circle. Surprisingly, even with rather large gaps there was significant suppression of brightness in the center of the target. We have encountered difficulties attempting to account for these findings with known physiological mechanisms such as lateral inhibition. A qualitative explanation of the results that looks promising is a two-component process involving brightness filling-in and smoothing to satisfy fixed boundary conditions at contours.