An anatomic analysis of retrograde and normograde intramedullary (IM) pinning of proximal, midshaft, and distal femoral fractures was performed in 28 canine cadavers. For all fracture locations, normograde pins were significantly more cranial in the middle gluteal muscle than retrograde pins (p less than 0.01). There was no significant difference between pinning techniques in craniocaudal position of the IM pin in the superficial gluteal muscle. In distal fractures, normograde pins were placed significantly more lateral than retrograde pins in the superficial gluteal muscle (p less than 0.01). One of 15 normograde pins and 9 of 13 retrograde pins were located in the medial half of the trochanteric fossa. Normograde pins were significantly more lateral in the trochanteric fossa than were retrograde pins in midshaft fractures (p less than 0.01). Normograde pins were significantly (p less than 0.01) farther from the sciatic nerve than retrograde pins when the hip was positioned at coxofemoral flexion angles of 85 degrees in midshaft and 110 degrees in distal fractures. Seven of 13 retrograde pins, but none of 15 normograde pins, contacted the sciatic nerve. Normograde pinning of the femur may be less likely to induce sciatic nerve injury, particularly in midshaft and distal fractures.