Data collected in the 1993 and 1994 cycles of an international External Quality Assessment (EQA) programme were cumulatively analysed to evaluate the analytical performance of the methods currently in use for routine assay of mucinous tumour markers CA 19-9, CA 15-3 and CA 125. On average the between-laboratory variability was 14.7 and 15.8 CV% for CA 15-3 and CA 125 respectively. For CA 19-9, a markedly worse between-laboratory variability (on average 27.2 CV%) was found; the agreement of CA 19-9 results worsened in the last few years when new isotopic techniques became available. The variability component attributable to systematic differences between methods/kits was relatively small for CA 15-3 and CA 125 (17% and 21% of the total variability), while it was markedly larger for CA 19-9 (45% of the total variability). The precision of the methods/kits most often used in the survey ranged from 9.6 to 13.9 CV% for CA 125 and from 10.8 to 14.1 CV% for CA 15-3. For these two tumour markers the precision of the traditional IRMAs does not appear to be different from that of the new fully automated non-isotopic techniques. The precision of CA 19-9 methods was on average worse from (11.9 to 19.2 CV%), even though the precision of the two automated systems was better than that of IRMAs. In conclusion, the results of this study indicate that the between-laboratory agreement for CA 15-3 and CA 125 assays appears satisfactory while the CA 19-9 assay shows larger differences between methods and is affected by poorer precision of kits.