Background: We evaluated the prognostic value of sequential NT-proBNP values in ambulatory heart failure patients after discharge, investigating whether the current value or the recent percent change is more important. Methods and results: In 166 patients, NT-proBNP was measured at discharge from heart failure hospitalisation and three months later. The combined endpoint of death or heart failure rehospitalisation was evaluated after a maximum of 18 months or at follow-up closure. During a mean observation time of 14 ± 4 months, 63 patients (38%) reached the endpoint. In multiple Cox analysis, NT-proBNP three months after discharge (NT-proBNP-3Mo) and NT-proBNP percent change (NT-proBNP-%change) were the only independent predictors of the endpoint among various clinical and laboratory variables. After definition of a high-(n = 83, 57% endpoints) and a low-NT-proBNP patient subgroup (n = 83, 19% endpoints) according to the median NT-proBNP-3Mo (1751 pg/ml), NT-proBNP-%change was the strongest predictor in the high-NT-proBNP subgroup. In the low-NT-proBNP subgroup, NT-proBNP-3Mo was the only independent predictor. Conclusions: In ambulatory heart failure patients, the prognostic value of sequential NT-proBNP measurements depends on the magnitude of the current NT-proBNP value. Recent percent changes in NT-proBNP provide important prognostic information in patients with high NTproBNP but not in patients with low NT-proBNP.