Clinical and standard radiographic evaluation of patients with lumbosacral radicular symptoms may, on occasion, fail to delineate a cause. This study retrospectively reviews 62 patients who had undergone nerve root infiltration (NRI) and assesses the accuracy and indications for this diagnostic study. Surgical exploration of patients with a Group 1 response (typical pain reproduced by needle placement and then relieved by NRI) confirmed local root pathology in all. Exclusive of patients with arachnoiditis, a Group 1 response showed 85% accuracy in identifying a single symptomatic root. A Group 2 response (typical pain reproduced by needle placement but not relieved by local anesthesia) indicated multiple root involvement. Patients with a Group 3 or Group 4 response (typical pain not reproduced by needle insertion, with or without relief of pain by local anesthesia) were seldom relieved of radicular pain. NRI was most useful in investigation of patients with radicular symptoms in whom other investigations were 1) normal, 2) showed multiple level involvement, or 3) were difficult to interpret because of previous surgery.