Introduction Revision for loosening of femoral stems requires an extensive analysis of bone defects to determine the most appropriate course of action. The drawbacks of using modular stems are that they can break or corrode at their junction. They have rarely been evaluated based on the extent of bone loss and particularly in patients with less severe bone loss. This led us to carry out a retrospective study to analyze modular femoral stems as a function of the initial bone defect (stage IIIB versus less severe in the Paprosky classification): 1) implant survivorship, 2) osteointegration and subsidence of the stem, and 3) breakage of implant. Hypothesis Modular femoral stems can be used for all types of bone defects (not only IIIB) as the complication rate is identical. Patients and methods Between January 1, 1996, and December 31, 2016, 163 patients were included who had received a modular femoral revision stem. The minimum follow-up was 4 years; the mean was 6.7 years ± 3.3 [4-21]. One patient was lost to follow-up, 88 had died before the analysis date and 74 were still alive; however, 10 of them had the stem removed less than 4 years after implantation. Thus 64 patients were available for the clinical evaluation. There were 44% (72 patients) with Paprosky stage IIIB femoral bone loss and 56% (91 patients) with stage I, II or IIIA bone loss. The stem's bone integration was evaluated using the Engh and Massin score. All complications were documented. Results The survivorship of the femoral stem was 93.75% (95% CI: 83.33–96.70) at 5 years with removal for any reason as the end point. There was no significant difference (p = 0.0877) in survivorship relative to the severity of the initial bone loss: 89.84% (95% CI: 78.73–95.31) for stage IIIB; 95.23% (95% CI: 82.24–98.79) for stage IIIA; 97.06% (95% CI: 80.90–99.58) for stage II. Bone integration was considered as being achieved in 76% of stems based on available radiographs (119 of 156 patients) with the severity of bone loss having no effect. We found 18 instances of stem subsidence out of 156 stems with available data (11.5%). The mean subsidence was 14.7 mm ± 12.3 [5-40]. Among the 18 stems with postoperative subsidence, 13 had been implanted for stage IIIB defects, while 5 were for less severe defects (p = 0.751). Two stem fractures occurred in patients with stage IIIB bone loss, thus 2/66 for stage IIIB and 0/86 in the less severe bone loss cases (p = 0.188). Conclusion Modularity provides similar results no matter the severity of initial bone loss, without the risk of additional complications. Level of evidence IV, Retrospective study.