ABSTRACT: da Silva, CD, Machado, G, Fernandes, AA, Teoldo, I, Pimenta, EM, Marins, JCB, and Garcia, ES. Muscle damage-based recovery strategies can be supported by predictive capacity of specific global positioning system accelerometry parameters immediately after soccer match-load. J Strength Cond Res 35(5): 1410-1418, 2021-Soccer match-load can be linked to recovery kinetic markers. However, match variability hinders the magnitude of relationship between parameters of interest. Therefore, we examined the correlation between 21 global positioning system accelerometry (GPS-A) parameters and changes in serum creatine kinase (CK) concentrations, muscle soreness (MS), and perceptive recovery quality (PRQ) assessed at baseline (1 h before) and post (0 minute, 2, 4, and 24 hours) a standardized 90-minute match-simulation in 20 university players. Global positioning system accelerometry (15 Hz) data were tested as manufacturer and configurable thresholds. Four GPS-A parameters showed moderate to very large correlations with CK changes at all time points (average speed [avgSP, r = 0.75 to r = 0.84]; running symmetry foot strikes [RSfst, r = 0.53-0.63]; running series [RunS, r = 0.53-0.61]; and acceleration distance [AccD ≥ 1.5 m·s-2; r = 0.46-0.61]). Sprint count (≥2 m·s-2), AccD (≥2.5 m·s-2) and speed exertion (SpEx) had a moderate to large correlation (r = 0.46-0.56) with CK changes from 2 to 24 hours. Changes in MS at 0 minute had large correlation with avgSP (r = 0.53) and moderate with deceleration distance (≥-2 and ≥-3 m·s-2; r = 0.47, r = 0.48, respectively). The PRQ changes had moderate inverse correlation with avgSP at 0 minute (r = -0.39) and SpEx at 2 h (r = -0.69). Our results suggest that during a simulated soccer protocol with a standard workload, only the avgSP has practical application for predicting CK changes over 24 hours, allowing for a decision-making toward a postgame recovery based on previously known CK cutoff points. Global positioning system accelerometry parameters and subjective variables did not demonstrate relevant correlation.