Background According to guidelines, the prognosis of patients with chronic heart failure can be predicted by determining the levels of natriuretic peptides, the NYHA classification and comorbidities. The aim our work was to develop a prognostic score in chronic heart failure patients that would take account of patients' comorbidities, NYHA and NT-proBNP levels. Methods and results A total of 1,088 patients with chronic heart failure with reduced ejection fraction (HFrEF) (LVEF<40%) and mid-range EF (HFmrEF) (LVEF 40-49%) were enrolled consecutively. Two-year all-cause mortality, heart transplantation and/or LVAD implantation were defined as the primary endpoint (EP). The occurrence of EP was 14.9% and grew with higher NYHA, namely 4.9% (NYHA I), 11.4% (NYHA II) and 27.8% (NYHA III-IV) (p<0.001). The occurrence of EP was 3%, 10% and 15-37% in patients with NT-proBNP levels <= 125 ng/L, 126-1000 ng/L and >1000 ng/L respectively. Discrimination abilities of NYHA and NT-proBNP were AUC 0.670 (p<0.001) and AUC 0.722 (p<0.001) respectively. The predictive value of the developed clinical model, which took account of older age, advanced heart failure (NYHA III+IV), anaemia, hyponatraemia, hyperuricaemia and being on a higher dose of furosemide (>40 mg daily) (AUC 0.773; p<0.001) was increased by adding the NT-proBNP level (AUC 0.790). Conclusion The use of prediction models in patients with chronic heart failure, namely those taking account of natriuretic peptides, should become a standard in routine clinical practice. It might contribute to a better identification of a high-risk group of patients in which more intense treatment needs to be considered, such as heart transplantation or LVAD implantation.