The influence of several methodological factors on MN counts revealed a more marked dependence mean values of sister chromatid exchanges (SCEs) on processing laboratories. This factor accounted and micronuclei (MN) in peripheral lymphocytes of for a percentage of roughly 10% (Pisa and Ca-Na) 1,650 subjects was analyzed. Donors belonged to of total variance, while the sampling period was a general healthy population living in Pisa and in marginally effective (about 1 -4% of total variabiltwo nearby small cities: Cascina and Navacchio ity). Because laboratories were equipped and sup-(Ca-Na). Blood samples were collected over a pe-plied with the same materials and consumables and riod of 29 months and processed in three different technicians were rotated constantly, the only varilaboratories of the same institute. Slides were ana-able ascertained was represented by the three diflyzed by several scorers. ferent models of CO 2 incubators used for lympho-Our data showed that lymphocyte proliferation cyte culturing. When ''month'' and ''incubator'' indexes (PIs) and baseline mean values of SCEs variables were considered jointly, experimental were affected mainly by sampling period. This fac-variability accounted for 15-20% of total variance, tor accounted for a percentage ranging from both for PIs and mean values SCEs and MN. The roughly 10% (Pisa) to 20% (Ca-Na) of total SCE variability due to slide scoring was reduced by asvariance and from roughly 10% (Pisa) to 13% (Ca-signing each slide to five different scorers and Na) of total PIs variance. A marginal effect was matching low with high scorers in each group. Presattributable to the different laboratories involved ent data show that when the study is performed (maximum 3% for SCEs and 7% for PIs). The sam-under these controlled conditions, about 20% of pling period variable included many sources of vari-total interdonor variability can be explained by exability such as culture media batches, fetal calf se-perimental or seasonal factors.