OBJECTIVES: To determine whether family physicians prefer discharge summaries in narrative or standardized format and to determine factors affecting this preference. DESIGN: Mailed survey. SETTING: Internal medicine ward at a teaching hospital. PARTICIPANTS: Random sample of 180 family physicians practising in the Ottawa-Carleton area. Of the original sample, 20 were not family physicians and were excluded. Of the 160 physicians remaining, 126 responded for a response rate of 78.8%. INTERVENTION: For a stratified random sample of patients, medical records and narrative discharge summaries were abstracted using a data acquisition form to capture essential information. Information on completed forms was transformed into standardized summaries. Physicians were sent both narrative and standardized summaries. MAIN OUTCOME MEASURE: Physicians' format preference as indicated on an ordinal 7-point scale. RESULTS: The standardized format was preferred with a score of 4.28 versus 3.84 for the narrative (P < .05). Responses indicated the standardized format provided information most relevant to ongoing care, with a mean score of 4.82 (95% confidence interval [CI] 4.48 to 5.15), and easier access to summary information (5.60, CI 5.30 to 5.89). The narrative summary better described patients' admission (3.54, CI 3.18 to 3.90). Preference for standardized summaries correlated with lengthier narrative summary (P < .05), shorter length of stay (P < .05), and physicians' dissatisfaction with previous summaries (P < .001). Standardized discharge summaries were significantly shorter (302 versus 619 words, P = .004) than narrative summaries. CONCLUSIONS: Physicians preferred a standardized format for discharge summaries. Format preference is influenced by physician, patient, and discharge summary characteristics.