Binding of [¹²⁵I] Tyr A¹⁴ human insulin ([¹²⁵I] insulin) was measured at 4°C in glomeruli and pieces of tubule microdissected from collagenase-treated rat kidneys. For glomeruli and all segments tested, total and non specific binding increased linearly with glomeruli number or tubular length. When determined with 4.0 nM labelled hormone, the distribution of specific binding sites (expressed as 10⁻¹⁸ mol [¹²⁵I] insulin bound per glomerulus or mm tubule length) was as follows: glomerulus, 2.5±0.3; proximal convoluted tubule (PCT), 12.6±0.6; pars recta (PR), 4.0±2.3; thin descending limb (TDL), 0.6±0.2; thin ascending limb (TAL), 0.6±0.2; medullary thick ascending limb (MAL), 0.8±0.1; cortical ascending limb (CAL), 2.1±0.1; distal convoluted tubule (DCT), 5.6±1.1; cortical collecting tubule (CCT), 3.2±0.3 and outer medullary collecting tubule (MCT), 2.3±0.1. Specific [¹²⁵I] insulin binding to glomeruli and tubule segments was time- and dose-dependent, saturable, reversible after elimination of free labelled ligand, and inhibited by unlabelled human insulin. When analysed in Scatchard and Hill coordinates, the binding data revealed a negative cooperation in the interaction processes between [¹²⁵I] insulin and glomerular and tubular binding sites, with apparent dissociation constants and Hill coefficients of the following values: glomerulus, 0.6 nM and 0.60; PCT, 10.0 nM and 0.55; MAL, 4.3 nM and 0.80; CAL, 2.0 nM and 0.74; CCT, 7.6 nM and 0.80 and MCT, 1.0 nM and 0.57 respectively. The stereospecificity of nephron binding sites was assessed in competitive experiments showing that unlabelled bovine and procine insulins were as efficient as human insulin for displacing [¹²⁵I] insulin, whereas A and B chains of insulin and unrelated peptide hormones were almost inactive. These results indicate that the detected [¹²⁵I] insulin binding sites may correspond to physiological insulin receptors.