This paper reports two studies of the effects of empathic instruction on first graders' evaluations of other people. In Experiment 1, 23 children received either empathic or non-empathic instructions, listened to a taped conversation in which the main character obtained either a positive or negative outcome, and then evaluated both the main character and the person who provided the outcome to the character. It was found that empathic instructions increased positive evaluation of the main character under negative outcome conditions only and produced a change in evaluative perspective in regard to both members of the interacting dyad. In Experiment II, 80 children received either empathic or neutral instructions, were informed that the main character was similar or dissimilar to themselves, and listened to a taped conversation in which the main character always obtained a mildly negative outcome. Female subjects evaluated the main character in the predicted sequence: empathy instructions and similarity information led to greatest liking, neutral instructions and dissimilarity information led to least liking, and the other two combinations led to intermediate levels of liking. Unexpectedly, male subjects responded to empathic instructions by evaluating the main character less favorably. Explanations for the males' evaluative behavior are discussed. Results are discussed in terms of whether empathic responses can be directly induced by instruction. (Author/RH)