During cancer monitoring, data on biological and analytical variation are required in order to define the critical difference which provides an objective means to interpret serial values. We evaluated four tumor markers on serial samples collected from healthy subjects and patients. Analytical coefficients of variation (CV(A)), were obtained from "precision profiles" based on the differences between duplicates cumulated from assay runs in the laboratory. We defined the mean intrasubject biological variation (CV(I)) for CA 19-9 and TPA, separately for healthy people and patients; since the differences between the two groups were not statistically significant, we pooled the results and re-evaluated CV(I) in the combined groups (CA 19-9: CV(I) = 15.9%; TPA: CV(I) = 25.7%). In addition, we evaluated CV(I) for CEA (10.9%) and for TPS (25.9%) in patients. We then evaluated the inter-subject biological variations (CV(G)); the calculated indices of individuality for the four markers were less than 0.6 which shows conventional reference values to be of little utility for interpretation. We finally evaluated the critical differences (p < 0.05) for CA 19-9 (CD = 44.7%), for TPA (CD = 72.5%), CEA (CD = 32.7%) and TPS (CD = 72.5%); these are generally applicable since there was no heterogeneity in intra-subject biological variability.