The present study evaluates optimal ventilatory criteria and exercise protocols for determining the ventilatory threshold, and assesses the day-to-day reproducibility of the ventilatory threshold and its relation to peak oxygen uptake VO2 and blood lactate concentration in normal subjects and patients with stable chronic congestive heart failure (CHF). Eighteen normal subjects and 18 patients underwent rapid (1-minute stage) and gradual (3-minute stage) bicycle exercise tests on consecutive days. The ventilatory threshold was determined from computer-generated printouts of expired gas variables measured breath by breath. Interobserver variability of ventilatory threshold was small in both normal (0.66 +/- 0.85 ml/min/kg) and CHF patients (0.50 +/- 0.46 ml/min/kg). Variability in the normal subjects was lower for the rapid exercise protocol (0.66 +/- 0.85 ml/min/kg) than the gradual protocol (1.72 +/- 1.63 ml/min/kg) (p less than 0.05), but both protocols produced similar results in the CHF group. Day-to-day reproducibility of ventilatory threshold was high (r = 0.91, standard error of the estimate 1.74 ml/min/kg) and was similar to that of peak VO2 (r = 0.95, standard error of the estimate 3.31 ml/min/kg). The use of co-plotted ventilatory equivalents for oxygen and carbon dioxide yielded ventilatory threshold values comparable to values obtained by using multiple parameters (r = 0.94, p less than 0.0001). Although the ventilatory threshold did not predict a precise lactate level for individual subjects, the lactate increment at the ventilatory threshold occurred within a narrow range in both normal subjects and patients with CHF; the increase was 7.5 +/- 4.5 mg/dl and 7.7 +/- 4.1 mg/dl, respectively, indicating a relation to initial increases in blood lactate.