Introduction In total hip replacement (THR), a short stem theoretically provides more physiological force transfer to the proximal femur, conserves bone stock and facilitates minimally invasive surgery. On the other hand, such implants involve a learning curve and incur risk of malpositioning or fracture and of secondary mobilization. There are several types of short stem, and classification is needed. Classifications exist, but are based more on implant length than on anchorage zone, and most have not been tested for reproducibility. The French Hip & Knee Society (SFHG) developed a short-stem classification based on anchorage zone inside the femur. The objectives of the present study were: (1) to present the classification, (2) to apply it to the short-stem models available in France and those widely used worldwide, and (3) to assess reproducibility. Hypothesis The SFHG short-stem classification enables reproducible comparison. Material and method A short-stem classification according to anchorage zone was drawn up by an expert group. The stems and the classification were presented to 12 surgeons performing THR, who classified the stems according to the classification; a retest was performed 2 months later. Results The classification is based on femoral stem anchorage site, in 5 types: type 1, cephalic; type 2, isolated cervical; type 3, Calcar femorale; type 4, metaphyseal; and type 5, conventional metaphyseal-diaphyseal, with shortened stems. Inter-observer reproducibility was 92.7% [95%CI: 91.7%–93.6%], with kappa 0.785 [95%CI: 0.755–0.814], and Lin test-rest concordance correlation coefficient 0.852 [95%CI: 0.836–0.869]. Intra-observer reproducibility was 94.0% [95%CI: 91.9%–96.1%], with kappa 0.820 [95%CI: 0.759–0.882], and Lin test-retest concordance correlation coefficient 0.820 [95%CI: 0.792–0.849]. Discussion This new classification enables femoral implants to be reproducibly compared according to anchorage zone. Level of evidence IV; retrospective study without control group.