Clonality studies have suggested that neoplasms are monoclonal and hyperplasias are polyclonal. To investigate this question in thyroid, we analyzed the clonality of 26 morphologically characterized hyperplastic nodules from 19 patients with sporadic goiters. For comparison we studied six thyroid carcinomas. We used the highly informative M27 beta probe that maps to the X-chromosome DXS255 locus (X cen-p11.22). Material was obtained from 52 nodules; tissue from nine nodules was rejected because of contamination with normal elements, five patients (eight nodules) were homozygous at Pst I sites in nonnodular thyroid tissue, and three nodules were excluded for technical reasons. Methylation patterns after Hpa II digestion confirmed polyclonality in all nontumorous thyroids of informative patients. Seven hyperplastic nodules were polyclonal, and 18 were monoclonal; one showed loss of heterozygosity. One nodule exhibited aberrant methylation. Multiple nodules were obtained from four patients; in three, all were monoclonal with activation of the same allele. Three papillary carcinomas were monoclonal; two exhibited aberrant methylation. One follicular carcinoma showed loss of heterozygosity. Our data indicate that morphologically indistinguishable hyperplastic thyroid nodules may be monoclonal or polyclonal. These findings suggest that variable molecular mechanisms are involved in the pathogenesis of nodules in sporadic goiter. Future studies will need to explore the biological significance of nodules of variable clonal origin.