This work analyses, in a first part, radioimmunoassay of serum digoxin levels in more than 4 500 patients during a period of 4 years using the same kit. A preliminary step of this work was to test the kits available 4 years ago i.e a kit using (3H) digoxin and 3 kits using (125I) digoxin. Results about this choice are summarized. The (3H) kit was finally chosen. It was used with slight modifications of the original protocol. This kit was as sensitive as 0.25 ng/ml and was specific for digoxin and its related derivatives. The reproducibility agrees with results in literature. Serum digoxin levels were assayed in the same samples, both with the (3H) kit and (125I) kits. Under these conditions, a lower average value was found with the former system. This result points out that the labelling protocol of the digoxin can modify the apparent normal therapeutic value observed when radioimmunoassay is done with different kits.