Gaze is a very important visual stimulus for survival and social interactions. 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 (Ando & Osaka, ARVO 1998). Since the human iris is usually darker than the white sclera, the human visual system may use this dark-light relation to compute the gaze direction efficiently. In order to examine this possibility, the luminance-induced gaze shift was first measured in photographic images of eyes. The effect was substantial: a moderate darkening of one side of the sclera induced an apparent shift of 8 to 10í of gaze; the maximum darkening induced a shift of 15í of gaze or more. The effect of scleral darkening was also compared to the gaze shift induced by an actual shift of the iris. Second, the effect was observed in schematic images of eyes. The two cues, scleral darkening and iris shift, had relative strengths similar to those measured in the first experiment. Predictions of the relative strengths of luminance and iris shift were developed for two simple luminance-based mechanisms: flux ratio and luminance centroid. The data showed that, compared to the physical shift of the iris, the luminance cue was less effective than predicted by either of those models. Third, the two cues were pitted against each other in a nulling paradigm in schematic images of eyes, and similar relative strengths were observed. Finally, irradiation could create a perceived shift in the iris position that might affect apparent gaze direction, however, a direct measure of the irradiation shift in a vernier task paradigm showed that it was too small by a factor of 10 compared with the effects measured in the second and the third experiment. The results suggest that one mechanism for gaze judgment is based on low-level analysis of the luminance configuration within the eye, but that it does not act alone.