The management of differentiated thyroid cancer in childhood is controversial. In particular, the role of aggressive surgical treatment has been questioned. This study was performed to identify those factors that are predictive of recurrence and morbidity following treatment through use of a multivariate model. The records of all patients 17 years of age or less admitted in the last 35 years with histologically confirmed differentiated thyroid carcinoma were reviewed. Data were sufficient for multivariate analysis in 93. The mean age at diagnosis was 13.3 years, and the median period of follow-up was 20 years. Seventy-one percent of the patients had nodal metastases. There were no deaths from thyroid carcinoma in this series, and the overall recurrence rate after initial treatment was 34%. Multivariate analysis demonstrated that only age (p less than or equal to 0.07) and histologic subtype (p less than or equal to 0.01) were predictive of time to recurrence. Major morbidity was a function of age (p less than or equal to 0.007) and extent of thyroid surgery (p less than or equal to 0.01). Probability of minor complications was predicted by use of radical neck dissection (p less than or equal to 0.02). Use of total or subtotal thyroidectomy or of radical neck dissection in children does not prevent recurrence and is associated with an increased risk of complications. We conclude that these procedures should be avoided in pediatric patients.