The medication prescribing practices on a University medical service were reviewed for a two year period. 103,484 medication orders were analyzed for 6,864 patient admissions. Forty drugs account for 64% of all medication orders. One hundred drugs account for 86% of all medication orders. During the two year period the incidence of prescribing specific drug combinations with potential adverse interactions varied from three to one hundred seventeen instances among the combinations examined. Review of a sample of those patients receiving the combination of spironolactone and oral potassium chloride showed that 52% developed hyperkalemia. These results suggest that when physicians prescribe drug combinations with potentially adverse interactions they may not always institute appropriate surveillance to prevent adverse consequences. A trial of automated monitoring of medication orders with appropriate physician feedback has begun as a result of these findings.