In patients with lupus nephropathy (LN), previous studies have shown that creatinine clearance (CCr) overestimates true glomerular filtration rate as measured by inulin clearance (CIn), and that among patients the degree of overestimation is highly variable. We sought to determine whether the discrepancy between CCr and CIn remains constant over time (months, years) in each individual patient, and therefore whether serial measurements of CCr reliably reflect the direction and magnitude of change in CIn. Twenty-five patients with LN underwent simultaneous determinations of CCr and CIn performed two to four (mean 3.3) times over three years. In a given patient, it was found that the ratio of CCr/CIn changed substantially over time (mean SD 0.16 with 95% confidence interval of 0.12 to 0.20). Thus, in about 32% of cases the ratio of CCr/CIn will vary more than +/- 16% from a previously measured value of CCr/CIn. Patients with both high and low values of CIn showed similar variability in CCR/CIn over time. Variability in CCr/CIn was found regardless of whether CIn was increasing, decreasing, or constant over time. In nearly one-half of all measurements of CCr, the corresponding change in CIn was directionally discordant. Iothalamate and technetium-DTPA renal clearances correlated highly with CIn (R2 = 0.99). We conclude that the discrepancy between CCr and CIn can vary greatly over time in an individual patient. Consequently, serial CCr does not accurately measure the direction or magnitude of change in glomerular filtration rate in lupus nephropathy.