Various combinations of modeling, labeling, and rehearsal (videotaped) were used to teach empathy to four experimental groups; a fifth group served as a no-treatment control. The subjects were 56 junior and senior baccalaureate nursing students, all of whom were women. A repeated-measures design was employed, with posttesting immediately following treatment and 3 weeks later. Learning was measured by means of a written test (Empathy Test) and an interview that was evaluated using the Barrett-Lennard Relationship Inventory and the Carkhuff Empathy Scale. The treatment was effective for junior students but not for senior students, and only the groups receiving the rehearsal conditioning performed better than the control group. There was also an interaction between treatment and time, with juniors improving on the second posttesting. A secondary hypothesis about correlations between the instruments was partially confirmed, thus lending support to their construct validity. The Carkhuff scale was correlated with itself for both testings, with the Barrett-Lennard inventory for both testings, and with the Empathy Test on the second testing.