Seventy-nine 3- and 4-year-old children were tested on gaze-reporting ability and Wellman and Liu's (2004) continuous measure of theory of mind (ToM). Children were better able to report where someone was looking when eye and head direction were provided as a cue compared with when only eye direction cues were provided. With the exception of younger boys, a significant number of children were able to perform at better-than-chance levels in both conditions. Children performed equally well with static and dynamic displays. A composite gaze-reporting measure was only weakly correlated with overall ToM; however, analyses of individual tasks revealed that children were more likely to pass a desire-reasoning task and fail gaze reporting than the reverse. The opposite was found for the relation between gaze reporting and the belief tasks. It was suggested that the ability to understand others' subjective desires, not the ability to understand others' representational, may be a precursor to having a "mentalistic" understanding of gaze. (Contains 4 tables.)