The biological and analytical components of variability of tumor markers should be distinguished from the variations due to tumor progression. The aim of the present study was to evaluate tumor marker variability in the follow-up of patients resected for breast cancer. So far, we have carried out 2,085 CEA and 2,550 CA 15-3 determinations in 435 patients. The total variability of both CEA and CA 15-3 was widely scattered among different subjects (CEA coefficient of variation 0-105%; CA 15-3 coefficient of variation 0-89.2%). The biological variability of CA 15-3, which was calculated in a limited number of cases, was scattered between 0 and 23% and was higher than the intra-assay variability. From these findings we conclude that when evaluating serial marker assays the intra-individual variability should be assayed initially to obtain a reference value of individual variability in relapse-free conditions.