AIMS: High-density lipoproteins (HDLs) are considered as anti-atherogenic. Recent experimental findings suggest that their biological properties can be modified in certain clinical conditions by accumulation of serum amyloid A (SAA). The effect of SAA on the association between HDL-cholesterol (HDL-C) and cardiovascular outcome remains unknown. METHODS AND RESULTS: We examined the association of SAA and HDL-C with mortality in the Ludwigshafen Risk and Cardiovascular Health (LURIC) study, which included 3310 patients undergoing coronary angiography. To validate our findings, we analysed 1255 participants of the German Diabetes and Dialysis study (4D) and 4027 participants of the Cooperative Health Research in the Region of Augsburg (KORA) S4 study. In LURIC, SAA concentrations predicted all-cause and cardiovascular mortality. In patients with low SAA, higher HDL-C was associated with lower all-cause and cardiovascular mortality. In contrast, in patients with high SAA, higher HDL-C was associated with increased all-cause and cardiovascular mortality, indicating that SAA indeed modifies the beneficial properties of HDL. We complemented these clinical observations by in vitro experiments, in which SAA impaired vascular functions of HDL. We further derived a formula for the simple calculation of the amount of biologically 'effective' HDL-C based on measured HDL-C and SAA from the LURIC study. In 4D and KORA S4 studies, we found that measured HDL-C was not associated with clinical outcomes, whereas calculated 'effective' HDL-C significantly predicted better outcome. CONCLUSION: The acute-phase protein SAA modifies the biological effects of HDL-C in several clinical conditions. The concomitant measurement of SAA is a simple, useful, and clinically applicable surrogate for the vascular functionality of HDL.