Although vouth workers frequently undergo in-service training, methodological shortcomings in evaluations of these programs often obscure the effects. This article reports on a field experiment that randomly assigned school counselors, juvenile court staff, and a variety of other youth workers to a training (experimental) or control group. Results support five of eight hypotheses comparing trainees with controls in terms of (1) tech niques such as reinforcement and (2) certain attributes of interpersonal relations with the youths in caseloads as well as with these youths' parents and friends. Shortcomings in the content of the training program appear responsible for inconclusive results on one hypothesis and lack of verification of two others.