This study examined the test-retest reliability, validity and external responsiveness of a new multi-change of direction agility test (NMAT) designed for soccer players. Forty-four Tunisian soccer players were recruited and were divided into two groups according to their playing levels (International, n = 21 and National, n = 23). Following familiarization, athletes performed squat jump (SJ), countermovement jump (CMJ), running speed test (5 m and 20 m), 15-m agility run (Agility-15 m), 15-m ball dribbling (Ball-15 m), and NMAT and ball dribbling NMAT (Ball-NMAT) in 2 sessions, 48 h apart. The intraclass correlation coef?cient and SEM values were .96 (CI 95%: .94 - .98) and .05 seconds for NMAT and .97 (CI 95%: .94 - .98) and .09 seconds for Ball-NMAT, respectively. The smallest worthwhile changes were greater than their SEM for both NMAT and Ball-NMAT. The MDC₉₅ values were .15 seconds and .25 seconds for NMAT and Ball-NMAT, respectively. Both NMAT and Ball-NMAT were respectively correlated with Agility-15 m (r = .78; p < .001) and Ball-15 m (r = .81; p < .001). Similarly, signi?cant correlations were observed between both NMAT and Ball-NMAT and leg power and straight sprint (.01 < p < .001). International-level soccer players were better than national-level in all tests including NMAT and Ball-NMAT (.01 < p < .001). The areas under their receiver operator characteristics curve were > .7 (.85; CI 95%: .71 - .94 and .91; CI 95%: .78 - .97 for NMAT and Ball-NMAT, respectively). These results indicated that NMAT provides excellent absolute and relative reliabilities. The NMAT can distinguish soccer athletes of different competitive levels. Thus, the NMAT may be suitable for field assessment of specific agility of soccer players.