We evaluated the analytical reliability of the carcinoembryonic antigen (CEA) method of Hansen. Our experience is based on performing over 23,000 CEA assays in more than 10,000 clinical samples. The sensitivity of the assay is 0.5 ng/ml. The precision must be defined as a function of concentration. In the range 2.6 to 12 ng/ml the coefficient of variation is 4.96 to 7.39%. Long-range studies of the reproducibility of the standard curve, over a period of several years and including 435 standard curves, have shown an overall mean B/Bo of 31.45% which is close to the theoretical optimal of 33%. The long-term 90 and 50% intercepts are 0.55 +/- 0.15 and 7.11 +/- 1.0, respectively. Interlaboratory surveys show good agreement between the means of the survey group and the target values but rather large individual discrepancies. The CEA method studied here is sensitive and reproducible in intralaboratory studies but less so in interlaboratory comparisons. The reagents have performed uniformly and close to specifications over an extended period of time.