The success of posterior lumbar interbody fusion has been analyzed primarily by plain radiography. There are no widely accepted criteria for fusion, except for lack of motion, which can be obviated by instrumentation, and the appearance of continuous bridging bone in the interspace. To determine the accuracy of plain radiography in detecting posterior lumbar interbody union, the radiographs of instrumented posterior lumbar fusion performed at 22 levels were compared with reformatted computed tomography (CT) scans. Results revealed that although 17 of 22 interbody levels were classified as fused by radiograph, 10 of these 17 demonstrated a nonunion by reformatted CT scans taken an average of 13.8 months postoperatively. Only progressively increased density by radiograph correlated with a solid fusion by CT scan. We therefore conclude that plain radiographs are unreliable in reporting the results of posterior lumbar interbody fusion and that CT scanning may be more appropriate.