OBJECTIVE: To determine the association of changes on nailfold capillaroscopy with clinical findings and genotype in children with juvenile dermatomyositis (DM), in order to identify potential differences in disease course over 36 months. METHODS: At diagnosis of juvenile DM in 61 children prior to the initiation of treatment, tumor necrosis factor alpha (TNFalpha) -308 allele and DQA1*0501 status was determined, juvenile DM Disease Activity Scores (DAS) were obtained, and nailfold capillaroscopy was performed. The disease course was monitored for 36 months. Variations within and between patients were assessed by regression analysis. RESULTS: At diagnosis, shorter duration of untreated disease (P = 0.05) and a lower juvenile DM skin DAS (P = 0.035) were associated with a unicyclic disease course. Over 36 months, end-row loop (ERL) regeneration was associated with lower skin DAS (P < 0.001) but not muscle DAS (P = 0.98); ERL regeneration and decreased bushy loops were associated with a shorter duration of untreated disease (P = 0.04 for both). At 36 months, increased ERL regeneration (P = 0.007) and improvement of skin DAS (P < 0.001) and muscle DAS (P = 0.025) were associated with a unicyclic disease course. CONCLUSION: Early treatment of juvenile DM may lead to a unicyclic disease course. The non-unicyclic disease course usually involves continuing skin manifestations with persistent nailfold capillaroscopy changes. The correlation of nailfold capillaroscopy results with cutaneous but not with musculoskeletal signs of juvenile DM over a 36-month period suggests that the cutaneous and muscle vasculopathies have different pathophysiologic mechanisms. These findings indicate that efforts to identify the optimal treatment of cutaneous features in juvenile DM require greater attention.