The importance of different factors in predicting post-traumatic growth was evaluated using multivariate statistical techniques. Forty-nine children with physeal fractures of the distal tibia or fibula or both were studied. The post-traumatic growth pattern was determined by roentgen stereophotogrammetry. The cases showed various types of growth patterns. The predictors used included sex, skeletal maturity, traumatological and anatomical classification, displacement, treatment, and associated tibial shaft fracture. Prediction of a specific growth pattern could be made only with low accuracy, whereas separation of the growth patterns into two groups--one group with symmetrical growth, initial and temporary growth retardation, and growth stimulation and the other with progressive growth retardation and growth arrest--resulted in a higher predictive accuracy. Comparing the predicted outcome with the actual resulted in two of three fractures being correctly classified. Considered individually, treatment, displacement and skeletal maturity were important, whereas the Salter-Harris classification system could not significantly predict the growth pattern. In summary, displaced physeal injuries treated by reduction showed the highest risk of developing deformity. These injuries were mainly combined fractures of the distal tibia and fibula with the physeal injury localized to one or both of the bones. The limited accuracy in predicting the post-traumatic growth pattern after ankle fractures in children indicates that in many cases there are grounds for radiographic follow-up to detect growth disturbance.