Acute renal failure (ARF) is associated with high mortality. Presently, no specific therapy for ARF exists. Therefore, early detection of ARF is critical to prevent its progression. However, serum creatinine, the standard marker to detect ARF, demonstrates major limitations. We prospectively evaluated whether serum cystatin C detected ARF earlier than serum creatinine. In 85 patients at high risk to develop ARF, serum creatinine and cystatin C were determined daily. ARF was defined according to the Risk of renal dysfunction, Injury to the kidney, Failure of kidney function, Loss of kidney function, and ESRD (RIFLE) classification when creatinine increased by >/=50% (R-criteria), by >/=100% (I-criteria), or by >/=200% (F-criteria). In analogy, ARF was detected when cystatin C increased by >/=50%, by >/=100%, or by >/=200%. Forty-four patients developed ARF and 41 served as controls. In ARF by R-, I-, and F-criteria, the increase of cystatin C significantly preceded that of creatinine. Specifically, serum cystatin C increased already by >/=50% 1.5 +/- 0.6 days earlier compared to creatinine. Serum cystatin C demonstrated a high diagnostic value to detect ARF as indicated by area under the curve of the ROC analysis of 0.82 and 0.97 on the two days before the R-criteria was fulfilled by creatinine. Cystatin C detected ARF according to the R-criteria with a sensitivity of 55% and 82% on these days, respectively. Cystatin C also performed excellently, detecting ARF defined by the I- and F-criteria two days prior to creatinine, and moderately well predicting renal replacement therapy in the further course of ARF. Additionally, low T(3)- or T(3)/T(4) syndrome, glucocorticoid deficiency and excess did not affect cystatin C levels, adding to its usefulness in critically ill patients with ARF. Serum cystatin C is a useful detection marker of ARF, and may detect ARF one to two days earlier than creatinine.