Chronic renal failure (CRF) is a common complication in heart transplant patients. Serum creatinine has clear limitations for the detection and estimation of glomerular filtration rate (GFR). Various creatinine-based formulae are classically used for GFR estimation, but little scientific evidence exists for such use in a heart transplant population. GFR was measured using the plasmatic clearance of the glomerular tracer (51)Cr-EDTA in 27 heart transplant patients with two measures for 22 of the patients. Forty-nine measures were thus available for analysis. The precision and accuracy (Bland and Altman analysis) of the Cockcroft, simplified Modified Diet in Renal Diseases (MDRD) and new Mayo Clinic formulae were compared. The mean GFR of the population was 39 +/- 15 mL/min/1.73 m(2). All formulae were well correlated with the GFR. With the Bland and Altman analysis, the accuracy of the MDRD formula appeared higher than that of the Cockcroft or the Mayo Clinic formulae (bias of +12 mL/min/1.73 m(2), vs. +19.9 mL/min/1.73 m(2), and +22.1 mL/min/1.73 m(2), respectively). The difference between the estimated and measured GFR was higher than 20 mL/min/1.73 m(2) in 51% and 55% cases when using the Cockcroft and the Mayo Clinic formulae respectively, whereas the difference was only noted in 14% cases when the MDRD was used. Among creatinine-based formulae, the MDRD appears the most precise and accurate for estimating the GFR in heart transplant patients. However, when the GFR must be measured with high accuracy, we recommend the use of a reference method like inulin or (51)Cr-EDTA plasma clearance techniques.