ABSTRACT The aims of this study were: (i) to analyse the load-velocity relationship in the bilateral leg-press exercise in female breast cancer survivors, (ii) to assess whether mean velocity (MV) or peak velocity (PV) show stronger relationship with the relative load, and (iii) to examine whether linear (LA) or polynomic (PA) adjustment predict the velocities associated with each %1RM with greater precision. Twenty-two female breast cancer survivors (age: 50.2 ± 10.8 years, weight: 69.6 ± 15.2 kg, height: 160.51 ± 5.25 cm) completed an incremental load test until 1RM in the bilateral leg-press exercise. The MV and the PV of the concentric phase were measured in each repetition using a linear velocity transducer, and were analysed by regression models using LA and PA. A very close relationship of MV (R 2 = 0.924; p < 0.0001; SEE = 0.08m.s−1 by LA, and R² = 0.952; p < 0.0001; SEE = 0.063 m.s−1 by PA) and PV (R² = 0.928; p < 0.0001; SEE = 0.119 m.s−1 by LA and R² = 0.941; p < 0.0001; SEE = 0.108 m.s−1 by PA) with %1RM were observed. The MV of 1RM was 0.24 ± 0.03 m·s−1, whereas the PV at 1RM was 0.60 ± 0.10 m.s−1. A comprehensive analysis of the bilateral leg-press load-velocity relationship in breast cancer survivors is presented. The results suggest that MV is the most recommendable velocity variable to prescribe the relative load during resistance training, and that the PA presents better accuracy to predict velocities associated with each %1RM, although LA is sufficiently valid to use this model as an alternative to the quadratic model. The implications for resistance training in breast cancer are discussed.