The aims of this study were to assess (i) the load–velocity relationship during the box squat exercise in women survivors of breast cancer, (ii) which velocity variable (mean velocity [MV], mean propulsive velocity [MPV], or peak velocity [PV]) shows stronger relationship with the relative load (%1RM), and (iii) which regression model (linear [LA] or polynomic [PA]) provides a greater fit for predicting the velocities associated with each %1RM. Nineteen women survivors of breast cancer (age: 53.2 ± 6.9 years, weight: 70.9 ± 13.1 kg, and height: 163.5 ± 7.4 cm) completed an incremental load test up to one‐repetition maximum in the box squat exercise. The MV, MPV, and the PV were measured during the concentric phase of each repetition with a linear velocity transducer. These measurements were analyzed by regression models using LA and PA. Strong correlations of MV with %1RM (R2 = 0.903/0.904; the standard error of the estimate (SEE) = 0.05 m.s−1 by LA/PA) and MPV (R2 = 0.900; SEE = 0.06 m.s−1 by LA and PA) were observed. In contrast, PV showed a weaker association with %1RM (R2 = 0.704; SEE = 0.15 m.s−1 by LA and PA). The MV and MPV of 1RM was 0.22 ± 0.04 m·s−1, whereas the PV at 1RM was 0.63 ± 0.18 m.s−1. These findings suggest that the use of MV to prescribe relative loads during resistance training, as well as LA and PA regression models, accurately predicted velocities for each %1RM. Assessing and prescribing resistance exercises during breast cancer rehabilitation can be facilitated through the monitoring of movement velocity.