Accurate empathy, long argued to be important in psychotherapy, now is an object of social-cognitive research. Graduate-level psychology students viewed brief portions of a therapy session and inferred the thoughts and feelings of the client. Accuracy scores were the rated similarity of their inferences to the client's reported thoughts and feelings. Throughout the semester course in interviewing, experimental participants practiced such judgments with feedback, while controls did not. Both groups' accuracy increased from pre-to post-test on inferred feelings, in part because the post-test was easier. Nonetheless, experimental participants on the post-test had greater accuracy of inferred feelings than controls. Women were more accurate than men in inferences for thoughts at post-test. Implications for training and future research are discussed.Empathy is a factor common to most theoretical orientations in clinical/counseling psychology as a means of increasing practitioners' effectiveness (Truax & Carkhuff, 1967). Empathy usually is conceptualized as a skill that can be developed and learned if properly instructed. The