To determine the mechanism responsible for the "placebo effect" seen during serial exercise testing of patients with heart failure, we examined metabolic variables for 81 patients who underwent five baseline exercise tests as part of a multicenter drug trial. The patients were 50 men and 31 women with a mean ejection fraction of 30.1% and a mean age of 69 years. From test 1 to 2, the exercise time increased from 419 +/- 140 to 462 +/- 130 seconds before it reached a plateau over the next three tests. Metabolic measurements at test 1 and test 3 revealed no change in peak oxygen consumption ( 1119 +/- 376 to 1105 +/- 346 ml/min). Maximum heart rate, systolic blood pressure, ventilation, and respiratory exchange ratio also were unchanged. The onset of the anaerobic threshold was delayed from 211 +/- 81 to 238 +/- 93 seconds, but there was no change in oxygen consumption at the anaerobic threshold (810 +/- 222 to 795 +/- 220 ml/min). At a predetermined submaximal level, oxygen consumption, ventilation, and respiratory exchange ratio all decreased to a statistically significant degree. These results indicate that a rapid increase in the mechanical efficiency of walking contributes to the placebo effect among patients with heart failure during serial exercise testing and is independent of changes in conditioning or motivation.