Data from a quality assessment scheme designed to test the measurement of cyclosporin are presented. Almost all of the participating 27 centres were using a radioimmunoassay for the measurement and, during the course of 16 months, a number of variables were tested, including accuracy, sensitivity, recovery of added cyclosporin, and within- and between-assay reproducibility. Accuracy of the measurement, judged by the mean results for spiked samples, was good, but values varied widely. Sensitivity tended to be overestimated, with a significant number of false-positive results being reported. The precision profile for the assay showed that there was no intrinsic difference in the precision of measurement using either plasma or whole blood as the sample matrix; but reproducibility for patient samples tended to be better using blood because drug concentrations are higher compared with plasma. We conclude that, whereas the results overall were encouraging, there were wide between-centre variations for the measurement, which could be compounded even further by the use of plasma or serum as the sample matrix.