The report by Strom et al1 on using computerized physician order entry to prevent the coadministration of warfarin and trimethoprim-sulfamethoxazole is an excellent example of the difficulty associated with developing usable computer-assisted drug interaction screening systems. However, it failed to address an important variable that often alters the risk of an adverse outcome resulting from a drug interaction. In the study, the intervention group received a hard-stop computer alert when a patient was prescribed the 2 drugs, regardless of the order of drug administration. It appears that the alert-triggering drug was warfarin in more than 90% of the cases in both control and intervention groups. That is, warfarin was being added to the drug regimen of patients already receiving trimethoprim-sulfamethoxazole.