The age-specific seroprevalence of varicella-zoster virus (VZV) antibodies was assessed in a sample of the Flemish (Belgian) population. ELISA tests were used to analyse 1673 sera from subjects aged 1 to 44 years (October 1999-April 2000). Chickenpox infections in Flanders appear to affect children at a younger age than in other European countries since 47.37% (95% CI: 37.33-57.41) is already immune at 2 years of age. For older age-groups, the prevalence is similar to that of most European countries: 80.19% (95% CI: 72.60-87.78) at 5 years, 92.52% (95% CI: 87.54-97.51) at 9 years and 100%> or =40 years. The accuracy of non-positive recollections of varicella histories among Flemish 10 to 17-year olds was examined on the basis of a second (residual) serum bank. In this group, VZV seroprevalence was almost always 100% (or nearly 100%), irrespective of age, degree of reliability (negative or uncertain answers) or level of ascertainment (child personally or parents). The limited size of this second data set did not allow for an accurate assessment of the negative predictive value of such recollections. CONCLUSIONS since varicella-zoster virus predominantly affects very small children and is generally perceived as benign, the required high coverage rate of a universal childhood varicella vaccination programme may be hard to attain. Adolescent strategies can minimise the population risks involved but the accuracy of non positive antecedents of chickenpox needs to be documented to assess the efficiency of such strategies.