Serum creatinine is an important clinical measure of impairment of glomerular filtration rate (GFR) after kidney transplantation. The use of formulas that predict GFR (such as the Cockcroft-Gault) derived from patients with chronic renal failure and standardized against measured creatinine clearance may not be accurate when applied to kidney transplant recipients. The purpose of this study, was to investigate the level of inaccuracy and its causes and then to derive predictive GFR formulas that are appropriate to renal transplantation. Determinants of isotopic GFR, serum creatinine, and muscle mass were evaluated in consecutive kidney recipients (n = 146) using 99mTc DTPA GFR (n = 751) as a reference method. Factors that predicted GFR apart from serum creatinine included sex, height, body weight, serum urea, years on dialysis, numbers of rejections and infective episodes, and prednisolone dose. The relationship between serum creatinine and GFR was highly variable and dependent on factors that alter muscle mass and muscle catabolic rate. The relationship was further altered by ATN and chronic rejection when tubular secretion of creatinine was reduced. Three alternative GFR formulas (which can be applied to renal transplant patients according to the availability of clinical parameters) were derived and tested against six published methods of GFR estimation. Our derived formulas had the highest correlation, no overall bias, least scatter of sum of squares, and least error at low levels of GFR. They represent a better estimation of GFR in kidney transplantation than published formulas, and would allow a standardized approach to the study of long-term renal dysfunction.