Detection of the ventilatory threshold during exercise has been proposed in order to assess exercise tolerance in patients with chronic heart failure. The relation between the different methods of detecting the ventilatory threshold and the lactate threshold, however, and their reproducibility, have not really been assessed. Forty-three patients with chronic heart failure underwent an exercise test with respiratory gas analysis. A lactate threshold could be determined in 36 patients and a ventilatory threshold in 27 to 38 patients, depending on the method of determination of the ventilatory threshold. The greatest number of determinations (38) and the best correlation coefficient with the lactate threshold (r = 0.87 and 0.88, respectively) were obtained with the method of the ventilatory equivalent for oxygen and by averaging the different methods of determination. Reproducibility of the ventilatory threshold was only moderately good (r = 0.83) and less satisfactory than that of the peak oxygen uptake (r = 0.97). We conclude that unless the way of detecting the ventilatory threshold is improved in patients with chronic heart failure, the peak oxygen uptake will remain more reproducible.