The most recent study by the Institute of Medicine of the reliability of hospital discharge data abstracted from patients' medical records documents the continuing presence of a substantial level of imprecision and error, affirming the findings of earlier studies. The most obvious factors leading to unreliability stem from inadequacies in the face sheets of medical records. Significant improvements could be attained if the discharge summary were routinely used to abstract information on patient disposition and principal diagnosis and if the operative report were routinely used to abstract information on principal procedure. Additional recommendations are offered for developmental activities intended to improve the designation of diagnosis, diagnostic classification schemes and hospital medical records systems.