In the evaluation of treatment failure in group A streptococcal pharyngitis, it is essential to distinguish persistence or relapse with homologous streptococcal strains from the acquisition of new, unrelated strains. Randomly amplified polymorphic DNA analysis and restriction fragment length polymorphism analysis of total DNA were used as epidemiological tools to compare 122 pre- and post-treatment Streptococcus pyogenes isolates obtained from 61 patients. The results obtained by molecular typing showed that bacteriological failures were due to the original strains in 43 cases (70%) and to new strains in 18 cases (30%). In the present study, restriction fragment length polymorphism analysis of total DNA appeared to be more discriminative than randomly amplified polymorphic DNA analysis. Thus, molecular analysis of DNA is an effective way to distinguish recurrence from persistence or relapse and will be useful in assessing the efficacy of new antibiotic treatments.