The aim of the study was to determine the extent to which glomerular filtration rate (GFR) measured with one filtration marker is associated with extracellular fluid volume (ECV) measured independently with another. Cr-51-EDTA and iohexol were injected simultaneously into opposite arms in 20 normal volunteers and 60 patients. Cubital samples taken bilaterally at 20, 40, 60, 120, 180 and 240 min were assayed for marker injected contralaterally. GFR was scaled to body surface area of 1.73 m(2) (GFR/BSA). GFR was also expressed as marker transit time (GFR/ECV) and ECV as the product of marker transit time and GFR/BSA. In normal subjects, changes between fasting and non-fasting ECV/BSA correlated positively with changes in GFR/BSA, but not GFR/ECV. GFR/BSA and GFR/ECV correlated positively (regression slope approximately 4 ml.min(-1).litre(-1)) and negatively (-2.7 ml x min(-1) x litre(-1)), respectively, with ECV/BSA. The difference, 6.7 ml x min(-1) x litre(-1), expressed as a fraction of average scaled GFR ( approximately 90 ml x min(-1)) is close to the reciprocal of average ECV/BSA (13.5 litres.1.73 m(-2)), consistent with the expected slope of the regression on ECV/BSA of the difference-to-average ratio of GFR/BSA and GFR/ECV. In contrast, in 29 patients with impaired GFR (estimated from plasma creatinine), ECV/BSA correlated inversely with GFR/ECV (slope approximately -5 ml x min(-1) x litre(-1)) but showed no relation with GFR/BSA. We conclude that in normal subjects GFR/BSA increases in response to increasing ECV/BSA, but the increase is not proportionate, leading to a weak inverse association between GFR/ECV and ECV/BSA. When ECV is expanded in patients with renal impairment, however, there is no GFR response, leading to a reduction in GFR/ECV.