To determine the frequency of ACE inhibitor cough in an outpatient medical clinic population, a cross-sectional epidemiologic survey using mailed questionnaires was done. Patients were randomly selected from a computerized hospital pharmacy data base. The overall prevalence of cough was 19 percent in the ACE inhibitor groups compared with 9 percent in the hydrochlorothiazide-treated group. The observed odds ratio for cough among ACE inhibitor users was 2.3 (95 percent CI, 1.02 to 5.00). This study is the first systematic investigation of frequency and characteristics of ACE inhibitor cough that includes a control group. Our results suggest that cough may more frequently accompany treatment with ACE inhibitors than has been previously reported. We recommend that physicians specifically inquire about cough in patients taking an ACE inhibitor. Recognition of this side effect may prevent unnecessary testing and treatment of patients receiving ACE inhibitors.