The outlook for children with Wilms' tumor has markedly improved with the use of multimodal therapy, and survival currently exceed 85%. Current trends have been to use less intense therapy for children whose tumors have favorable histology (FH). By decreasing the amount of therapy given to patients whose conditions have an excellent chance of responding, the need for accurate criteria to separate high-risk and low-risk groups becomes imperative. Nuclear morphometric techniques have been developed and extensively tested in the author's laboratory. Preliminary studies of FH Wilms' tumors demonstrated the applicability of this technique. Herein, the authors present a retrospective study of 108 patients with FH tumors, with a mean follow-up period of 1,994 +/- 107 days (66 months) for those whose tumors did respond to therapy, and 686 +/- 61 days (23 months) for those whose tumors did not respond. The univariate predictors were age (P = .02), the skewness of nuclear roundness factor (SNRF) (P = .009), and the mean of the lower five values for nuclear ellipticity, measured by the feret diameter method (L5EFD) (P = .01). A multivariate analysis combining all three variables better separated the two groups (P = .00016). A probability function curve was constructed to predict poor prognosis. Kaplan-Meier actuarial survival analysis was used to predict disease-free survival as clinical progression, and separated the two groups (P = .0004, Wilcoxon-Gehan statistic). These results suggest that nuclear morphometry is useful in the initial assessment of patients with Wilms' tumor.