In patients with fluid retention, the plasma clearance of (51) Cr-EDTA (Clexp obtained by multiexponential fit) may overestimate the glomerular filtration rate (GFR). The present study was undertaken to compare a gamma-variate plasma clearance (Clgv) with the urinary plasma clearance of (51) Cr-EDTA (Clu ) in patients with cirrhosis with and without fluid retention. A total of 81 patients with cirrhosis (22 without fluid retention, 59 with ascites) received a quantitative intravenous injection of (51) Cr-EDTA followed by plasma and quantitative urinary samples for 5 h. Clgv was determined from the injected dose relative to the plasma concentration-time area, obtained by a gamma-variate iterative fit. Clexp and Clu were determined by standard technique. In patients without fluid retention, Clgv , Clexp and Clu were closely similar. The difference between Clgv and Clu (Clgv - Clu = ΔCl) was mean -0·6 ml min(-1) 1·73 m(-2) . In patients with ascites, ΔCl was significantly higher (11·8 ml min(-1) 1·73 m(-2) , P<0·0001), but this value was lower than Clexp - Clu (17·5 mL min(-1) 1·73 m(-2) , P<0·01). ΔCl increased with lower values of GFR (P<0·001). In conclusion, in patients with fluid retention and ascites Clgv and Clexp overestimates GFR substantially, but the overestimation is smaller with Clgv . Although Clu may underestimate GFR slightly, patients with ascites should collect urine quantitatively to obtain a reliable measurement of GFR.