Of several methods for constructing a radioimmunoassay (RIA) standard curve discussed in this paper the logit-log transformation emerges as the most favourable approach. In our hands, all RIA systems analyzed so far display a linear dose-response relationship after logit-log transformation. Truncating the standard curve to its operative range between effective dose 85 and 15, linear regression analysis can also be performed with the unweighted data, thus considerably facilitating computation. A standardized concept for RIA evaluation and quality control has been developed which can be executed by a low-cost desk-top calculator. A standardized protocol for RIA analyses should include the determination of total radioactivity added, the amount of both non-specifically and maximally-bound tracer hormone, 6 logarithmically-diluted standards specially prepared to cover the range between ED 85 and 15 and control samples with different levels of analyte. The RIA standard curve should be described by appropriate numerical parameters to keep a record of the intrinsic assay performance quality. Within assay variance is found on the basis of dose scatter of all unknown run in replicate, while between assay variance is based upon the observation of control sample variability. Control charts should graphically exhibit the variation of both essential assay parameters and control samples. Quality control should mainly be regarded as a means to assist in assay optimization which obviously is an important prerequisite for international standardization of RIA analyses.