A new method for the immunoassay of insulin using guinea-pig anti-human insulin serum, ¹³¹I-labelled ox insulin, and rabbit anti-γ-globulin serum is described. Three methods (A, B, and C) are outlined. Method A uses anti-insulin serum to isolate insulin from a fixed amount of ¹³¹I-labelled insulin mixed with standard or unknown unlabelled insulin. The insulin-antibody complex is precipitated with anti-γ-globulin serum and separated by filtration. Radioactivity in the precipitate is measured to determine insulin concentration using a standard curve. Method B involves precipitating insulin antibody with anti-γ-globulin serum and using the precipitate to sample insulin mixtures. Method C is a modification of B, where precipitated insulin antibody is incubated with unlabelled insulin before adding ¹³¹I-labelled insulin, increasing sensitivity. Plasma γ-globulin can interfere in method A if anti-γ-globulin serum is not in excess, but not in methods B or C. Methods B and C are preferred for insulin assay in blood plasma. Methods A and B are based on isotope dilution, but in practice, the amount of insulin bound by antibody varies with insulin concentration. With ¹³¹I-labelled insulin of specific activity 5–20 mc/mg., 6 × 10⁻⁶ i.u. (6 microunits) of human insulin/ml. can be detected. The plasma insulin concentration in five normal people was (in microunits/ml.): after starvation, 16; and 30, 60 and 150 min. after the oral administration of glucose (50 or 100 g.), 64 and 158, 65 and 49, and 22 and 21 respectively. The method is accurate and superior to biological assays. It is applicable to the specific microassay of any protein that can be purified and labelled with a suitable radioactive isotope, is antigenic, and where precipitation of the antibody with anti-γ-globulin serum does not interfere with antigen binding.A new method for the immunoassay of insulin using guinea-pig anti-human insulin serum, ¹³¹I-labelled ox insulin, and rabbit anti-γ-globulin serum is described. Three methods (A, B, and C) are outlined. Method A uses anti-insulin serum to isolate insulin from a fixed amount of ¹³¹I-labelled insulin mixed with standard or unknown unlabelled insulin. The insulin-antibody complex is precipitated with anti-γ-globulin serum and separated by filtration. Radioactivity in the precipitate is measured to determine insulin concentration using a standard curve. Method B involves precipitating insulin antibody with anti-γ-globulin serum and using the precipitate to sample insulin mixtures. Method C is a modification of B, where precipitated insulin antibody is incubated with unlabelled insulin before adding ¹³¹I-labelled insulin, increasing sensitivity. Plasma γ-globulin can interfere in method A if anti-γ-globulin serum is not in excess, but not in methods B or C. Methods B and C are preferred for insulin assay in blood plasma. Methods A and B are based on isotope dilution, but in practice, the amount of insulin bound by antibody varies with insulin concentration. With ¹³¹I-labelled insulin of specific activity 5–20 mc/mg., 6 × 10⁻⁶ i.u. (6 microunits) of human insulin/ml. can be detected. The plasma insulin concentration in five normal people was (in microunits/ml.): after starvation, 16; and 30, 60 and 150 min. after the oral administration of glucose (50 or 100 g.), 64 and 158, 65 and 49, and 22 and 21 respectively. The method is accurate and superior to biological assays. It is applicable to the specific microassay of any protein that can be purified and labelled with a suitable radioactive isotope, is antigenic, and where precipitation of the antibody with anti-γ-globulin serum does not interfere with antigen binding.