Background: Exogenous tracer-based methods of measuring glomerular filtration rate (GFR) are difficult to perform, whilst creatinine-based estimation formulae are inaccurate. Methods: We assessed a new technique of measuring iohexol clearance using timed dried capillary blood spots. A reference GFR was measured in 81 subjects (GFR 15–124 ml/min/1.73 m<sup>2</sup>) by iohexol clearance using three venous samples (2, 3 and 4 h after an intravenous bolus). GFR was estimated by six test methods; iohexol clearance using (i) 3 blood spots (2, 3, 4 h); (ii) 2 blood spots (2, 4 h) and (iii) 1 blood spot (4 h); (iv) the Modification of Diet in Renal Disease (MDRD) formula; (v) the Cockcroft-Gault formula, and (vi) a formula estimating GFR from serum cystatin C concentration. For each test method the bias and precision were calculated as the mean and standard deviation (SD) of the ‘GFR differences’ (test method GFR – reference GFR). Results: The limits of agreement (bias ±1.96 × SD; in ml/min/1.73 m<sup>2</sup>) were: (i) 1.1 ± 15.1 for 3-spot iohexol clearance; (ii) 0.6 ± 14.9 for 2-spot iohexol clearance; (iii) 4.5 ± 21.2 for 1-spot iohexol clearance; (iv) –15.7 ± 33.3 for the MDRD formula; (v) –9.6 ± 32.9 for the Cockcroft-Gault formula, and (vi) –12.1 ± 31.7 for the Cystatin C formula. The accuracy of all six test methods was similar among individuals with GFR <60 ml/min/ 1.73 m<sup>2</sup>; however, in individuals with GFR ≧60 ml/min/ 1.73 m<sup>2</sup>, the MDRD, Cockcroft-Gault and Cystatin C formulae were all imprecise and systematically underestimated GFR. Conclusions: Blood spot iohexol clearance provides a potentially practical method of estimating GFR accurately in large-scale epidemiological studies especially among individuals without established chronic kidney disease.