The aims of this study were to evaluate the reliability of a modified agility T-test (MAT) and to examine its relationship to the free countermovement jump (FCMJ) and the 10-m straight sprint (10mSS). In this new version, we preserved the same nature of displacement of the T-test but we reduced the total distance to cover. A total of 86 subjects (34 women: age = 22.6 +/- 1.4 years; weight = 63.7 +/- 10.2 kg; height = 1.65 +/- 0.05 m; body mass index = 23.3 +/- 3.3 kg x m(-2) and 52 men: age = 22.4 +/- 1.5 years; weight = 68.7 +/- 8.0 kg; height = 1.77 +/- 0.06 m; body mass index = 22.0 +/- 2.0 kg x m(-2)) performed MAT, T-test, FCMJ, and 10mSS. Our results showed no difference between test-retest MAT scores. Intraclass reliability of the MAT was greater than 0.90 across the trials (0.92 and 0.95 for women and men, respectively). The mean difference (bias) +/- the 95% limits of agreement was 0.03 +/- 0.37 seconds for women and 0.03 +/- 0.33 seconds for men. MAT was correlated to the T-test (r = 0.79, p < 0.001 and r = 0.75, p < 0.001 for women and men, respectively). Significant correlations were found between both MAT and FCMJ, and MAT and 10mSS for women (r = -0.47, p < 0.01 and r = 0.34, p < 0.05, respectively). No significant correlations were found between MAT and all other tests for men. These results indicate that MAT is a reliable test to assess agility. The weak relationship between MAT and strength and straight speed suggests that agility requires other determinants of performance as coordination. Considering that field sports generally include sprints with change direction over short distance, MAT seems to be more specific than the T-test when assessing agility.