This study investigated the predictability of maximal oxygen intake from three different submaximal heart rates assessed during an initial and follow-up ride on a cycle ergometer. Twenty-four healthy male subjects performed workloads of 600, 750, and 900 kpm's for six minutes on each of two visits to the laboratory. Analysis of variance for a randomised complete blocks design, with subjects constituting blocks, was used to analyse heart rate, estimated maximal oxygen intake, and residual estimated maximal oxygen intake variations among the experimental conditions. Relationships between the actual and estimated maximal oxygen intakes were determined using the Pearson Product-Moment formula of correlation. The average estimated maximal oxygen intake was significantly increased from the first testing occasion to the second. Although errors of estimation decreased significantly (450 ml to 366 ml) in favour of the second testing condition, the decrease was neither consistent with workloads nor subjects. The correlation coefficients were consistently low at 600 kpm for both testing occasions (0.68 and 0.73, respectively), consistently high at 750 kpm (0.82 and 0.84, respectively), and quite variable at 900 kpm (0.71 and 0.84, respectively), indicated that the validity of the nomogram was not consistent with all workloads or testing occasions. Despite these inconsistencies, the nomogram is, for practical purposes, a valid predictor of maximal oxygen consumption.