This study aimed to compare skating force-velocity relationships determined throughout sprints performed against various loaded conditions or inferred from movement kinetics measured during a single unloaded sprint. Ten female ice hockey players performed one unloaded maximal skating sprint test measured with a radar gun followed by 4 resisted skating sprints against a robotic horizontal resistance with progressive loads in reference to equipped body mass (BM): 3 kg (robotic resistance), 25, 50, and 75% of equipped BM. Maximal theoretical force (F0), velocity (V0), power (Pmax), optimal velocity (Vopt) condition for producing maximal power, and slope of the linear force-velocity relationship (SFV) were determined from each method and compared using a paired sample t-test, absolute mean bias (±95% confidence intervals), Pearson correlations, and typical error of the estimate in standardized units (effect size [ES]). Statistical significance was set at p < 0.05. No statistical difference was found for all mechanical variables determined from the 2 methods (p ranging 0.09–0.59). Although exhibiting positive correlations ranging from moderate (r = 0.50 for SFV) to high (r ranging from 0.71 to 0.84 for F0, V0, Vopt, and Pmax) between methods, all variables exhibited large levels of error between approaches (ES ranging 0.66–1.71). Multiple loaded and single unloaded methods were comparable with determine force-velocity relationships during forward on-ice skating sprint. The low-cost fatigue-free unloaded method suggests it could be used in constrained contexts (i.e., congested schedule and low available time) or for a simple force-velocity profiling. Inversely, multiple loaded methods would be more appropriate to evaluate and individualize training for skilled ice hockey players accustomed to resistive skating sprint.