The study aim was to determine whether urine albumin/creatinine ratio (UACR), high-sensitivity C-reactive protein (hsCRP) or N-terminal pro-brain natriuretic peptide (Nt-proBNP) added to risk prediction based on HeartScore and history of diabetes or cardiovascular disease. A Danish population sample of 2460 individuals was divided in three groups: 472 subjects receiving cardiovascular medication or having history of diabetes, prior myocardial infarction or stroke, 559 high-risk subjects with a 10-year risk of cardiovascular death above 5% as estimated by HeartScore, and 1429 low-moderate risk subjects with estimated risk below 5%. During the following 9.5 years the composite end point of cardiovascular death, non-fatal myocardial infarction or stroke (CEP) occurred in 204 subjects. CEP was predicted in all three groups by UACR (HRs: 2.1, 2.1 and 2.3 per 10-fold increase, all P<0.001) or by hsCRP (HRs: 1.9, 1.9 and 1.7 per 10-fold increase, all P<0.05), but not by Nt-proBNP (HRs: 1.1, 2.6 and 3.7 per 10-fold increase, last two P<0.001) (P<0.05 for interaction). In the low-moderate risk group, pre-specified gender adjusted (men/women) cutoff values of UACR> or =0.73/1.06 mg mmol(-1) or hsCRP> or =6.0/7.3 mg l(-1) identified a subgroup of 16% who experienced one-third of the CEPs. In the patient group, combined absence of high UACR and high Nt-proBNP> or =110/164 pg ml(-1) (men/women) identified a subgroup of 52% who experienced only 15% of the CEPs. Additional use of UACR and hsCRP in subjects with low-moderate risk and UACR and Nt-proBNP in subjects with known diabetes of cardiovascular disease changed HeartScore risk classification significantly in 19% of the population.