Recently, measurement of serum et-fetoprotein (sAFP) was introduced as a preliminary test for diagnosis of Fanconi's anemia (FA). In the present study, sAFP levels were measured in order to determine its sensitivity and specificity in 33 Turkish FA patients(17 males and 16 females) with a mean age of 11.6 +/- 7.7 (1.0-28.0) (median 10.0). Complementation groups were available in 12 patients. Nineteen age-matched healthy children, 17 patients with bone marrow failure syndromes, 37 FA heterozygotes, and 37 children with acute leukemia served as negative control groups. The sAFP was measured by particle immunoassay. The level of sAFP was found to be higher than the cut-off value, 8 IU/mL in 46% and was within normal limits in 54% of the FA patients. The AFP values were within normal limits in all of the subjects belonging to the control groups. This method provided 46% sensitivity and 100% specificity in the diagnosis of FA. The sAFP values were high in 4 of 17 (24%) FA patients who did not receive any androgen therapy, while the sAFP level was high in 7 of 9 (78%) patients who received such a therapy. The statistical analysis of incidence of a high sAFP level between these two groups indicated a significant difference (P = 0.014), suggesting that androgen therapy might be a contributing factor for elevation of sAFP. The comparison of several clinical and laboratory parameters between FA patients with high and normal levels of AFP revealed no statistically significant differences. The level of sAFP was elevated in only 5 of the 11 patients with complementation group A; in addition, variable levels of sAFP were noted among the affected members in 4 families, indicating that complementation groups, type of mutation, or familial factors were not responsible for elevation of sAFP.