Twenty isolates belonging to nine Fusarium spp. were tested for levels of pathogenicity on two cotton cultivars under greenhouse conditions. Preemergence damping-off, postemergence damping-off, survival and dry weight were used as criteria to evaluate pathogenicity of the isolates. Proteins and alcohol dehydrogenase (AD) isozymes of the isolates were separated by SDS-PAGE, and PAGE, respectively. Virulence patterns, protein banding patterns, and AD isozymic patterns patterns were used to differentiate among Fusarium isolates belonging to the same species. In the three methods, cluster analysis was used to differentiate among the isolates, which showed low similarity levels-that is, the isolates, which belonged to remotely related or unrelated subclusters. On the other hand, cluster analysis was not a reliable method to differentiate among the isolates belonging to closely related subclusters because such isolates had relatively high similarity levels. The results of the present study suggest that isolates belonging to the same Fusarium species could be differentiated by their differential pathogenicity, combined with their specific protein and AD banding patterns separated by electrophoresis.