This investigation was undertaken to identify clinical variables, alone or in combination, that could be used to assign children to high- and low-risk categories for intra-abdominal injury following blunt trauma. Six hundred consecutive children who were examined with computed tomography (CT) following blunt trauma were enrolled. Complete data sets were available on 375 children. Stepwise logistic regression was used to identify predictor variables for the presence of abdominal injury. There were 174 children with abdominal injury detected by CT. Of these, 95 were classified as having significant injury. Indicators associated with significantly higher risk of abdominal injury included the following: more than three clinical indications given (odds likelihood ratio [OLR] = 4.60, 95% confidence interval [95% Cl] = 2.29, 9.21, p less than 0.001); gross hematuria (OLR = 5.80, 95% Cl = 2.51, 13.4, p less than 0.001); lap belt injury (OLR = 12.2, 95% Cl = 2.22, 66.8, p less than 0.01); assault or abuse as the mechanism of injury (OLR = 5.08, 95% Cl = 1.07, 24.2, p less than 0.05); abdominal tenderness (OLR = 2.73, 95% Cl = 1.296, 5.82, p less than 0.01); and Trauma Score less than or equal to 12 (OLR = 2.27, 95% Cl = 1.006, 5.13, p less than 0.01). No child with asymptomatic hematuria (n = 56), regardless of grade or neurologic impairment in the absence of abdominal findings (n = 15), had an abnormal CT examination. These data are useful as an adjunct to clinical judgment in triage when the availability of CT equipment is limited or there are competing extra-abdominal injuries.