Myotonic dystrophy type 1 (DM1) is associated with an expansion of CTG repeats in the 3'UTR of the DMPK gene. It is accepted, as in other trinucleotide diseases, that the number of the repeats is correlated with age at onset and severity of the disease. However, assessment of genotype-phenotype correlation in DM1 is complicated with the expansion-biased somatic instability of mutant alleles over time and difficulties in precise assessment of the number of repeats by standard Southern blot hybridization. In order to clarify this issue we defined DM1 expansion size in lymphocytes by three parameters: size of progenitor, average, and largest allele, using a more precise small-pool/long-range PCR technique. We found a negative linear correlation of age at onset and average expansion size in juvenile-adult DM1 patients (35 out of 46) whose progenitor allele is less than 245 repeats long. Our result favors the hypothesis of the existence of a threshold in the progenitor allele size beyond which number of CTG repeats does not influence age at onset. Potential clinical significance is that the average allele size could be a useful indicator for the age at onset in juvenile-adult DM1 patients with relatively short progenitor allele. To test whether somatic instability of mutant alleles influences the progression of DM1, patients were divided in three phenotypic classes according to the severity of neuromuscular symptoms. We showed that the largest expansion in each DM1 phenotypic class reflects somatic instability of mutant allele over time independently of progenitor allele size and patient's age at sampling. The mean of the largest expansion was significantly different between phenotypic classes, implying the possible association between expansion-biased somatic instability of mutant alleles over time and progression of neuromuscular symptoms.