BACKGROUND Serum levels of cystatin C have been proposed to be an ideal marker of the glomerular filtration rate (GFR). However, some reports have shown that serum levels of cystatin C increase independently of GFR. In this study, we evaluated the clinical utility of cystatin C in monitoring GFR, especially in patients with a malignancy. METHODS Study subjects consisted of 82 patients with a malignancy, 39 patients with a non-malignancy, 31 healthy volunteers, and 206 patients with various degrees of renal function. We measured serum cystatin C, beta2-microglobulin (beta 2mG), and creatinine (CRE) levels in all patients. Serum CRP levels were measured in 21 patients with a malignancy and 28 patients with a non-malignancy whose creatinine clearance (Ccr) was > or =70 ml/min. Cystatin C, beta 2mG, and CRP were measured by immune nephelometry and CRE was measured by an enzyme assay. RESULTS In patients with a malignancy, regression analysis yielded the equation: 1/cystatin C = 0.06 x Ccr + 0.710, correlation coefficient, r, of 0.33. The r was significantly lower than in patients with various degrees of renal function. There were no significant differences when the r performed on beta 2mG and CRE was compared between the same groups of patients. In 74 patients with a malignancy, in whom serum CRE levels were < or =1.1 mg/dl, increased levels of cystatin C were observed in 25 patients and increased levels of beta 2mG were observed in 39 patients. In comparing patients with a malignancy and a non-malignancy, the number of patients with an increased level of cystatin C, despite a Ccr > or = 70 ml/min (8/33) or a CRE < or = 1.1 mg/dl (13/41), was larger in the former group than the latter group, although the result was not statistically significant. Similarly, the number of patients with an increased level of beta 2mG, despite a Ccr > or = 70 ml/min or a CRE < or = 1.1 mg/dl was significantly larger in the former group compared to the latter group. Regression analysis between the serum levels of cystatin C and CRP in patients with a malignancy whose Ccr were > or =70 ml/min had a weak correlation (r = 0.31). CONCLUSIONS The results of our study suggest that the serum levels of cystatin C are not always a reliable marker of the GFR in patients with a malignancy, probably in relation to its nature as a cysteine protease inhibitor.