OBJECTIVE Our goals were to determine (1) whether Korotkoff phase IV or V sound was a more accurate measure of diastolic blood pressure in pregnancy and (2) interobserver variability of mercury sphygmomanometry of pregnant women. METHODS Direct (intraarterial) and indirect (mercury sphygmomanometry) blood pressures were compared in 28 pregnant women. Interobserver variability was assessed in a separate study of 86 pregnant women using four highly trained observers. RESULTS (1) Routine sphygmomanometry underestimated direct systolic pressure by 11 (3, 18) mm Hg, p < 0.001 (median, interquartile range of differences). Phase IV Korotkoff sound overestimated direct diastolic pressure by 9 (2, 12) mm Hg (p < 0.001) and phase V by 4 (2, 7) mm Hg (p = 0.04). Phase V-recorded diastolic pressure was closer to direct diastolic pressure significantly more often (75%) than was phase IV-recorded diastolic pressure (21%) (p = 0.003). Mean arterial pressures did not differ significantly according to the method used. (2) Median blood pressures did not differ among the four observers for systolic, diastolic phase IV, or phase V recordings. Maximum difference for blood pressure recording among observers was 4 (2, 6) mm Hg. CONCLUSIONS Auscultatory sphygmomanometry in pregnant women underestimates systolic and overestimates diastolic blood pressure, but the phase V Korotkoff sound is more likely to represent the true diastolic pressure than is the phase IV sound.