Samples from two pools were sent 10 times to 27 laboratories for assay of digoxin. One pool contained digoxin 2.60 nmol/l in normal plasma (SP); the other was pooled plasma from patients treated with digoxin (PP). Ten radioimmunoassay (RIA) methods were used. The mean of SP assays was 2.59 nmol/l, not significantly different from 2.60 nmol/l. The mean of PP determinations was 2.46 nmol/l. Within each of the 10 assay rounds, the concentrations showed an almost twofold variation and S.D. averaged 0.33 nmol/l and 0.31 nmol/l for SP and PP respectively. Significant differences (P less than 0.001) were found between mean concentrations obtained for the pools at various laboratories (SP range 2.15-2.85 nmol/l; PP range 2.12-2.72 nmol/l). The laboratory means obtained for SP and PP correlated significantly (P less than 0.001). Nevertheless, significant (P less than 0.01) variations between laboratories were found also concerning the mean difference between SP and PP concentrations. The interassay SD of the assays differed significantly between laboratories (range 0.05-0.61 nmol/l. Between and within groups of laboratories using the same RIA method and between various types of laboratories, differences were also found concerning both accuracy and precision of the assays. It is concluded that a better control of digoxin assay is needed.