Femoral intramedullary canal referencing is utilized by most of the total knee arthroplasty (TKA) systems. Violation of the canal is performed in order to engage rod instruments in the femoral diaphysis and to refer of the anatomical axis of the femur. Fat embolism, activation of the coagulation cascade, and bleeding may occur from the reamed femoral canal. The purpose of our study was to validate a new set of _minimally-invasive friendly_ instruments which allow to prepare the femur without violating the intramedullary canal. Twenty-five consecutive patients undergoing primary TKA through a mini-subvastus approach were enrolled in the study after informed consent had been obtained. Results of this cohort (group 1) were compared to another contemporary group (group 2) of 25 TKAs operated by the same surgeon using intramedullary instruments. The two groups were matched for gender, deformity, degree of arthritis, and surgical approach. Reliability of the new extramedullary set of instruments was first tested in ten cadaveric limbs. Preoperative long weight-bearing AP and lateral view of the knee were obtained taking care of neutral limb positioning. Template of the mechanical and anatomical axis were performed. Distal femoral resection was planned according to the template, and considering a bone cut perpendicular to the mechanical axis of the femur. Measurement from the template were reproduced on the distal femoral cutting jig. Flexion-extension control of the distal femoral resection was obtained using the anterior meta-diaphyseal cortex reference. Depth of resection, and varus-valgus angulation were selected according to the previous measurements and referring over the most prominent distal femoral condyle. A double check was performed using an extra-medullary rod referring two and a half finger-breaths medially to the antero-superior iliac spine. Postoperative blood loss, pain, swelling, functional recovery, and complications were recorded. Radiographic alignment was measured with long film. Mechanical axis was within 0±2° in 88% of group 1 and 84% of group 2 (p> 0.05). There were no difference between the two groups regarding the operative time. In group 1, postoperative blood loss (740 vs 820 mL) was reduced but this difference did not reach the statistical significance (p=0.07). No difference was found in terms of postoperative pain, knee swelling, and functional recovery. Extramedullary reference with careful preoperative templating can be safely utilized during total knee arthroplasty. Avoiding the violation of the femoral canal may enhance the benefits of a less invasive approach.