The ability to assess psychosocial problems of a group of primary care internal medicine residents education in a relatively unstructured psychiatric liaison program was compared to a group of internists who had no specialized psychiatric training. The primary care residents did not demonstrate a superior ability to assess patients' psychosocial problems when compared to internists without such training. The effectiveness of such unstructured liaison experiences (i.e., the Balint model of group supervision and teaching conferences) is questioned. The authors suggest that the main benefit of such informal approaches may lie in gaining initial attitudinal and administrative acceptance by nonpsychiatric physicians of liaison teaching programs. Formally structured, supervised learning experiences are probably necessary to effectively enhance internists' clinical skills and behaviors in dealing with the psychosocial and psychiatric aspects of medical care.