PURPOSE: Treating constipation in children with voiding dysfunction may improve or resolve urinary symptoms. A clinical diagnosis of constipation may not identify all patients. Abdominal radiographs (plain x-ray of the kidneys, ureters and bladder) are often used to assess constipation but no objective definition of constipation based on abdominal radiographs exists. Most abdominal radiograph rating scales use subjective criteria and our previous series showed that these scales have poor reliability. We identified reliable, objective parameters on abdominal radiograph to predict constipation. MATERIALS AND METHODS: The abdominal radiographs of 80 children 4 to 12 years old, including 40 with constipation and 40 successfully treated for constipation, were assessed for several measurable parameters. Logistic regression was used to construct a model to predict constipation status based on these abdominal radiograph parameters. Model accuracy was assessed using AUC analysis of ROC curves. RESULTS: The most predictive model included cecal diameter, total length of stool measured, stool length in the rectum, and patient age and gender. As measured by the area under the ROC curve, accuracy was excellent at 0.87. We calculated cutoffs for individual parameters on abdominal radiograph, including total stool length greater than 33.4 cm, cecal diameter greater than 3.7 cm and stool length in the rectum greater than 5.9 cm. CONCLUSIONS: We identified accurate, reliable criteria based on objective measurements on abdominal radiograph to differentiate patients with and without constipation. These criteria may be applied to objectively assess constipation status in children with urinary symptoms without a history of constipation. Further study will determine whether these criteria predict the response to treatment.