The purpose of this study was to identify factors that predict implant cut-out after cephalomedullary nailing of intertrochanteric and subtrochanteric hip fractures, and to test the significance of calcar referenced tip-apex distance (CalTAD) as a predictor for cut-out. We retrospectively reviewed 170 consecutive fractures that had undergone cephalomedullary nailing. Of these, 77 met the inclusion criteria of a non-pathological fracture with a minimum of 80 days radiological follow-up (mean 408 days; 81 days to 4.9 years). The overall cut-out rate was 13% (10/77). The significant parameters in the univariate analysis were tip-apex distance (TAD) (p <  0.001), CalTAD (p = 0.001), cervical angle difference (p = 0.004), and lag screw placement in the anteroposterior (AP) view (Parker’s ratio index) (p = 0.003). Non-significant parameters were age (p = 0.325), gender (p = 1.000), fracture side (p = 0.507), fracture type (AO classification) (p = 0.381), Singh Osteoporosis Index (p = 0.575), lag screw placement in the lateral view (p = 0.123), and reduction quality (modified Baumgaertner’s method) (p = 0.575). In the multivariate analysis, CalTAD was the only significant measurement (p = 0.001). CalTAD had almost perfect inter-observer reliability (interclass correlation coefficient (ICC) 0.901). Our data provide the first reported clinical evidence that CalTAD is a predictor of cut-out. The finding of CalTAD as the only significant parameter in the multivariate analysis, along with the univariate significance of Parker’s ratio index in the AP view, suggest that inferior placement of the lag screw is preferable to reduce the rate of cut-out. Cite this article: Bone Joint J 2014; 96-B:1029–34