We studied the frequency of ethical problems in a general medical ward at a university hospital, using a quasi-experimental prospective design. In the baseline period, ethical problems were determined by self-report of residents. In the intervention period, one of us (B.L.) was a participant-observer during attending rounds. Ethical problems were determined by consensus among the attending physician, resident, and participant observer. No significant differences between baseline and intervention periods were found in patient variations, admissions per resident, or rating of each resident's sensitivity to ethical issues. In the baseline period, seven (3.9%) of 179 cases involved ethical problems. In the intervention period, 16 (17%) of 92 cases involved ethical problems. This difference was significant. The data imply that residents underidentify ethical problems but that sensitization, sympathetic listening, information, and advice increase physician recognition of ethical problems.