Starting from 1984 we organized an interlaboratory quality control (QC) survey for the tumor marker AFP in which 100-150 laboratories have been involved. Seven consecutive QC cycles have been carried out during the period 1984-1988; 15-20 samples have been prepared and sent in each cycle. The 11 kits more used in the survey, which use three different immunological methods (3 kits EIA, 3 kits RIA, and 5 kits IRMA) were evaluated. Since the results of AFP assays were expressed as ng/ml of different local standards of the various kits, the participants were asked to convert their results in UI/ml of the 1st IRP 72/225 distributed by WHO. The total variability of AFP assay resulted approximately constant (19.9-24.7 CV%) during the all QC period. The validity of the consensus mean of AFP determinations obtained in the survey and assumed as reference value has been checked by recovery experiments. Regression analysis indicates that the consensus means found in these samples are in very good agreement with the added AFP. The analytical performance achieved by the most popular methods/kits, observed in the last period, are also reported.