The major problems in applying quality control to radioimmunoassay measurements are (a) nonlinearity of the dose-response curve, and (b) nonuniformity of the residual variance. A logit-log transformation of the dose-response variables combined with an iterative weighted regression analysis appears to overcome most of the difficulties. This technic is particularly helpful when applied to substandard runs where extraneous standard points tend to distort assay results. The authors describe a quality control program that involves recording control values on calendar and histogram formats, monitoring assay variables by charting, and comparing the computer-calculated slope with the graphic plot to reveal "outliers." This program is useful in guiding technologists to locate possible causes for "out-of-limits" runs.