To compare unreamed intramedullary nailing (IMN) with external fixation (EF) in patients with Type II, IIIA, and IIIB open fractures of the tibial shaft. An inception cohort of consecutive patients with Type II, IIIA, and IIIB tibial fractures incurred between January 1988 and March 1993 were systematically allocated into one of two treatment groups. Patients were treated and followed with a prospectively designed protocol. All patients were skeletally mature and had incurred a fracture of the tibial diaphysis within twenty-four hours of presentation to the tertiary care hospital, a Level I Trauma Center. One hundred seventy-four fractures in 168 patients were stabilized with either IMN (104) or half-pin EF (70). There were 132 men and thirty-six women, with an average age of thirty-three years (range, 14 to 77 years). Except for the selection of the fixation device, open fracture care was similar in the two treatment groups. All patients underwent emergent irrigation and debridement with concomitant skeletal stabilization. Cephalosporin antibiotics were administered perioperatively for twenty-four to forty-eight hours. No wounds were closed primarily. Delayed primary closure, skin grafting, and/or myoplasty were performed between three and ten days after injury. The main outcome measures were final fracture alignment, presence of infection or inflammation, hardware failure, time to union, and the number of operative procedures. The IMN group had significantly fewer incidences of malalignment than did the EF group [8 vs. 31 percent; p = 0.00005; confidence interval (CI) = 0.18, 0.76] and had significantly fewer subsequent procedures (mean of 1.7 vs. mean of 2.7 per fracture; p = 0.001; CI = 0.45, 1.59). IMN resulted in fewer infections/ inflammatory problems than did EF at the injury site (13 vs. 21 percent; p = 0.73; CI = -0.63, 0.45) and significantly fewer at surgical interfaces (i.e., pin sites, nail and interlocking screw insertion sites; 2 vs. 50 percent; p = 0.000; CI = 0.39, 0.60). No significant difference was found in the healing rates for the two implant groups. The more severe Gustilo injury types had longer healing times regardless of the type of fixation. Results suggest that unreamed interlocking intramedullary nails are more efficacious than half-pin external fixators, in particular with regard to maintenance of limb alignment. However, the severity of soft tissue injury rather than the choice of implant appears to be the predominant factor influencing rapidity of bone healing and rate of injury site infection.