Aims To analyse measures of clinical data, functional capacity, left ventricular function and neurohormonal activation for the ability to predict mortality and morbidity in patients after a hospitalisation for heart failure. Methods In a prospective study, patients 60 years or above with systolic heart failure NYHA II–IV were followed for at least 18 months. At study start, a physical examination, echocardiography, blood samples and measurements of quality of life (QoL) by Nottingham Health Profile were obtained. Data on mortality and readmission rates were collected. Results 208 patients, 58% men, with a mean age of 76 years, and an ejection fraction of 0.34 were included and followed for a mean of 1122 days. In all, 74 (36%) patients died and 171 (82%) were readmitted. By univariate analysis, readmissions were predicted by poor QoL (169 ± 118 vs. 83 ± 100, p < 0.001), age, creatinine, haemoglobin ( p < 0.01 all) and diabetes ( p < 0.1). By multivariate analyses, QoL at study start was the only independent predictor of readmissions ( χ 2 = 25.2, p < 0. 001). Mortality was univariately associated with QoL (183 ± 117 vs. 142 ± 115, p < 0.05) and in multivariate analyses to traditional variables: age, male gender, systolic function, BNP and serum creatinine ( χ 2 = 48.9, p < 0.001). Conclusions Measurements representing different aspects of the heart failure syndrome can easily be obtained to stratify long-term risks of mortality and morbidity in hospitalised heart failure patients. Poor QoL was a univariate predictor for mortality and a strong multivariate predictor for the important outcome of readmission, pointing to the need for a simple assessment of QoL.