Using probit analysis, dose-response curves for induction of anesthesia with midazolam or ketamine were constructed in ASA class III and IV patients premedicated with morphine, 0.1 mg/kg, and glycopyrrolate, 4 micrograms/kg. For ketamine, ED50 values for abolition of the response to verbal commands, eyelash stimulation, and painful stimulation were 0.9, 1.3, and 1.3 mg/kg, respectively; corresponding ED95 values were 1.6, 2.3, and 4.3 mg/kg, which are within the range of clinically recommended doses. For midazolam, ED50 values for verbal commands, eyelash stimulation, and painful stimulation were 0.19, 0.24, and 0.36 mg/kg, significantly greater than those previously reported for unpremedicated ASA class I and II patients. The corresponding ED95 values, 0.35, 0.43, and 1.04 mg/kg exceed previously reported values and are appreciably greater than the doses used in most previous studies of midazolam induction. Midazolam decreased systolic blood pressure slightly but significantly (from 138 +/- 4 to 128 +/- 4 mm Hg, mean +/- SEM, P less than 0.005), while diastolic blood pressure and heart rate remained unchanged. In contrast, ketamine increased systolic blood pressure (from 141 +/- 4 to 164 +/- 5 mm Hg, P less than 0.005), diastolic blood pressure (from 71 +/- 3 to 88 +/- 4 mm Hg, P less than 0.005), and heart rate (from 84 +/- 2 to 102 +/- 4 beats/min, P less than 0.005). On the basis of these data, we conclude that in ASA class III and IV patients, midazolam induction allows for hemodynamic stability and avoids the significant tachycardia and hypertension associated with equipotent doses of ketamine.