Changing the luminance of one side of the sclera induces an apparent shift of the perceived direction of gaze toward the darker side of the sclera. This luminance-induced gaze shift was measured in photographic and schematic images of eyes. The effect was substantial: a moderate darkening of one side of the sclera induced an apparent shift of 8 to 10 deg of gaze; the maximum darkening induced a shift of 15 deg of gaze or more. The effect of scleral darkening was also compared to the gaze shift induced by an actual shift of the iris. The effects of the two cues were measured independently and in combination. When pitted against each other, their effects could be nulled, demonstrating that they act on a common level. Predictions of the relative strengths of the luminance and iris shift cues were developed for two simple luminance-based mechanisms: flux ratio and luminance centroid. The data showed the luminance cue was less effective than the models predicted in determining gaze direction. As an alternative source for the gaze shift, irradiation effects on apparent size could create a perceived shift in the iris position but a direct measure of the irradiation shift showed that this was far too small. The results suggest that at least one important mechanism for gaze judgment is based on low-level analysis of the luminance configuration within the eye.