OBJECTIVE To evaluate the reliability of voiding cystourethrography (VCUG) to diagnose infravesical obstruction in boys. METHODS Hard copies of the VCUG findings of 72 boys were assessed by two pediatric radiologists and two pediatric urologists. The investigators were instructed to consider six items related to infravesical obstruction: vesicoureteral reflux, bladder wall thickness, bladder diverticulum, abnormal prostatic urethra, visible urethral obstruction, and obstruction in the sphincter area. Also, a scoring system was developed using these six items. Agreement among the four investigators for evaluation of the items on VCUG and for the scoring system was assessed using the kappa statistic. All boys underwent urethrocystoscopy, and the endoscopic findings were compared with VCUG results. Odds ratios were calculated for the results of VCUG for each investigator to predict the chance of cystoscopic infravesical obstruction. RESULTS Agreement among observers for vesicoureteral reflux and bladder diverticulum was good (kappa values for paired observers of 0.82 and 0.79). Agreement for bladder wall thickness, abnormal prostatic urethra, visible urethral obstruction, obstruction in the sphincter area, and the scoring system was poor (kappa values of 0.08, 0.35, 0.33, 0.26, and 0.33, respectively). Consequently, the results of VCUG could not predict for endoscopic infravesical obstruction accurately, although substantial differences occurred among investigators. Items on which investigators reached good agreement were negatively related to the risk of having infravesical obstruction. CONCLUSIONS Because agreement among investigators in the assessment of most items on VCUG was poor and because for the items with good agreement, the predictive power was poor, the current clinical use of VCUG for diagnosing infravesical obstruction needs reevaluation.