The application of data-processing methods such as deconvolution or computer-assisted background subtraction to renography requires knowledge of the input to the kidneys. The two most commonly used approximations to the kidney input in 131I-hippuran probe renography are to assume that the kidney input is described either by the time-activity curve recorded over the subclavian region (mode I) or by the time derivative of mode I (mode II). A third approximation to the kidney input using two externally recorded time-activity curves, pre-injection of 131I-albumin and two venous samples has previously been described. This approach has been modified in this work, in which two externally recorded time-activity curves and two capillary samples are used to construct an estimate for the kidney input (mode III). The three modes are compared with a reference, which is constructed from capillary samples drawn during the renography examination. Eleven patients participated in this study, all of whom were referred for routine renography. The results show that only mode III is close to the reference in all cases.