OBJECTIVE The aims of this study were to determine the sensitivity of voiding cystourethrography (VCUG) in detecting incontinence in women with a history of leakage, the positive and negative predictive values of VCUG for stress incontinence, and how often a history of leakage is accompanied by urethrocele on VCUG and to correlate urethral anatomic measurements with the presence of urethrocele and stress incontinence on VCUG. METHODS A total of 159 women with incontinence or voiding dysfunction were evaluated by VCUG and urodynamic study (UDS). RESULTS VCUG detected stress incontinence in 76% of women with a specific history of stress incontinence. Of 61 women with genuine stress incontinence proven on UDS, only 37 (61%) were identified on VCUG, yielding a positive predictive value of 56%. Twenty-one of 29 women without genuine stress incontinence on UDS but with apparent stress incontinence on VCUG had detrusor instability on UDS, so that urge incontinence was falsely diagnosed as stress incontinence. VCUG had a negative predictive value of 74%. In 40 women with urethroceles, 26 had stress incontinence on VCUG. The anatomic changes in posterior urethrovesical angle, urethral descent, and urethral inclination denote urethroceles but do not correlate with the presence of stress incontinence. CONCLUSIONS In the evaluation of stress incontinence, VCUG is limited because of detrusor instability producing urge incontinence and therefore resulting in false-positive stress incontinence. The anatomic measurements of posterior urethrovesical angle change, urethral descent, and urethral inclination as well as the presence of urethrocele have limited ability in predicting stress incontinence.