Anaerobic threshold was studied in 35 patients with cardiac failure present for more than 6 months, of primary or ischemic origin and New York Heart Association Stage II or III. Each patient underwent an exercise test of ergometric bicycle (increments of 10 watts/1 minute) with analysis of expired gases respiratory cycle by respiratory cycle (oxygen consumption, carbon dioxide production, respiratory oxygen and carbon dioxide equivalent, respiratory quotient). Anaerobic threshold was determined indirectly following Wasserman's recommendations. It was possible to determine anaerobic threshold in 25 patients (83.3%). The test was well tolerated. The two best methods for detection of anaerobic threshold were divergence of the curve of production of carbon dioxide in relation to oxygen consumption and an increase in respiratory oxygen equivalent while the carbon dioxide equivalent remained constant. In addition, oxygen consumption at the anaerobic threshold was statistically correlated with functional class (p < 0.05) and with the Weber classification based upon peak oxygen consumption (p < 0.001). Finally, anaerobic threshold occurred on average at 5.4 +/- 1.8 min from the start of exercise with a heart rate of 119.7 +/- 18 bpm, an oxygen consumption of 10.5 +/- 2.3 ml/min/kg (i.e. 43.2 +/- 12.1% of maximum oxygen consumption and 70.5 +/- 19.3% of oxygen consumption limited by symptoms) and a respiratory quotient of 1.07 +/- 0.08. In conclusion, analysis of respiratory gases during exercise in cardiac failure patients enables, with sub-maximal exercise and noninvasively, the determination in the majority of cases of anaerobic threshold, which is statistically correlated with New York Heart Association stage and the Weber classification based upon oxygen consumption at peak effort.