Recently, in a cross-sectional study, a correlation of moderate degree was documented between serum BNP (brain natriuretic peptide) and exercise capacity in patients with chronic heart failure (CHF). However, it remains unknown if BNP, which increases in response to high myocardial wall stress, is sufficiently sensitive for changes in exercise capacity during clinical follow-up. To elucidate this, 42 CHF patients were recruited and randomized into a training (T; 58 +/- 10 years; n = 14 NYHA II; n = 5 NYHA III) and a control group (CO; 54 +/- 9, n = 17 NYHA II; n = 6 NYHA III). T carried out 12 weeks of endurance training on a cycle ergometer (4 sessions per week, 45 min duration). Venous blood sampling and cycle ergometry with simultaneous gas exchange measurements were carried out prior to and after the experimental phase. Due to its superior stability during laboratory procedures, NTproBNP was determined instead of BNP. Both proteins are secreted in equimolar amounts and share an identical diagnostic meaning. In both groups, NT-proBNP decreased slightly (T: from 1092 +/- 980 to 805 +/- 724 pg x ml(-1); CO: from 1075 +/- 1068 to 857 +/- 1138 pg x ml(-1); T vs CO: p = 0.65). Anaerobic threshold (AT) as a measure of exercise capacity went up in T (from 0.96 +/- 0.17 to 1.10 +/- 0.22 l x min(-1)) but remained almost constant in CO (pre: 1.02 +/- 0.27; post: 1.00 +/- 0.27 l x min(-1); T vs CO: p < 0.001). The correlation between changes in NT-proBNP and changes in AT remained insignificant (r = 0.02, p = 0.89)-even if only T was considered (r = 0.09, p = 0.72). Improved exercise capacity in CHF patients due to 3 months of endurance training is not reflected in the course of NT-proBNP. These findings are inconsistent with a sufficient sensitivity of this parameter to detect changes in exercise capacity during clinical follow-up. Changes in NT-proBNP beyond its spontaneous variability are more likely to be detected following therapeutical interventions which aim more clearly at the myocardium. In determining alterations of functional capacity ergometric testing cannot be replaced by serial determinations of NT-proBNP.