A dural tear resulting in a cerebrospinal fluid (CSF) leak is a well-known risk of lumbar spinal procedures. The authors hypothesized that the incidence of CSF leakage is higher in cases involving repeated operations and those in which the surgeon performing the surgery is less experienced; however, they postulated that the overall outcome of the patient would not be adversely affected by a dural tear. An institutional review board-approved protocol at Louisiana State University Health Sciences Center, Shreveport, was initiated in August 2003 to allow prospective comparison of data obtained in patients in whom a CSF leak occurred (Group A) and those in whom no CSF leak occurred (Group B) during lumbar surgery. Basic demographic information, descriptive findings regarding the tear, history of other surgeries, hospital length of stay (LOS), and immediate disposition at the time of discharge were compared between the two groups. Seventy-seven patients were eligible for this study. One patient refused to participate. In 12 (15.8%) of 76 patients CSF leakage developed. In three patients the presence of a tear was questioned, and the patients were clinically treated as if a tear were present. The patients in Group A were older than those in Group B (59.8 +/- 16.9 and 49.4 +/- 13.6 years of age, respectively; p = 0.02, Fisher exact test). In terms of those with a history of surgery, there was no significant difference between patients with and patients without a CSF leak (three [25%] of 12 patients [Group A] compared with 28 [43.8%] of 64 patients [Group B]; p = 0.34, two-sample t-test). In the 12 patients with dural tears, nine (75%) were caused by a resident-in-training, and the Kerrison punch was the instrument most often being used at the time (55%). This is significantly greater than 50% at the 5% level (p = 0.044, binomial test). The authors were able to repair the tear primarily with suture in all but one patient, whose tear was along the nerve root sleeve. In all cases fibrin glue and a muscle/fat graft were used to cover the tear, and all patients were assigned to bed rest from 24 to 48 hours after the operation. In Group A one patient required rehabilitation at discharge. The LOS in Group A was greater than that in Group B (median 5 days compared with 3 days), but no additional complication was noted. The incidence of CSF leakage was 16% in 76 patients, and there were no other complications. Older patient age and higher level of the surgeon's training were factors contributing to the incidence, but the history of surgery was not.