We measured maximal O2 uptake (VO2max) during stationary cycling in 40 pregnant women [aged 29.2 +/- 3.9 (SD) yr, gestational age 25.9 +/- 3.3 wk]. Data from 30 of these women were used to develop an equation to predict the percent VO2max from submaximal heart rates. This equation and the submaximal VO2 were used to predict VO2max in the remaining 10 women. The accuracy of VO2max values estimated by this procedure was compared with values predicted by two popular methods: the Astrand nomogram and the VO2 vs. heart rate (VO2-HR) curve. VO2max values estimated by the derived equation method in the 10 validation subjects were only 3.7 +/- 12.2% higher than actual values (P greater than 0.05). The Astrand method overestimated VO2max by 9.0 +/- 19.4% (P greater than 0.05), whereas the VO2-HR curve method underestimated VO2max by only 1.6 +/- 10.3% in the same 10 subjects (P greater than 0.05). Both the Astrand and the VO2-HR curve methods correlated well with the actual values when all 40 subjects were considered (r = 0.77 and 0.85, respectively), but the VO2-HR curve method had a lower SE of prediction than the Astrand method (8.7 vs. 10.4%). In a comparison group of 10 nonpregnant sedentary women (29.9 +/- 4.5 yr), an equation relating %VO2max to HR nearly identical to that obtained in the pregnant women was found, suggesting that pregnancy does not alter this relationship. We conclude that extrapolating the VO2-HR curve to an estimated maximal HR is the most accurate method of predicting VO2max in pregnant women.