To determine whether anticipation in myotonic dystrophy (DM) is a true biologic phenomenon or an artifact of ascertainment bias, we studied 201 members of nine DM kindreds, including 67 individuals with the clinical diagnosis of DM. Of 49 parent-child pairs in which both the parents and the children were clinically affected, the onset of DM occurred in an earlier decade of life in the child than the parent in 44 pairs and in the same decade in five pairs (p < 0.001). To eliminate direct ascertainment bias, we excluded nine pairs involving the index patients. Indirect ascertainment bias due to incomplete penetrance was unlikely, since 55% of the children of DM parents had DM. However, by haplotype analysis of restriction fragment length polymorphisms, we diagnosed DM in one of the 42 asymptomatic children of affected parents and excluded DM in twenty-eight. We estimated that patients with early-onset DM would have produced an additional 25 DM children if they had normal fertility and nuptiality. Assuming that the expected age-of-onset distribution occurs without anticipation in these 25, only seven would have had the onset of DM earlier than their parents. With the corrected result, the child would have been affected earlier than the parent in 53 pairs, and the parent would have been affected at the same age as or earlier than the child in 13 pairs (p < 0.001). Thus, the observed anticipation is unlikely to be totally attributable to ascertainment bias, suggesting the potential importance of biologic mechanisms.