Purpose: The author tried to examine the water and sand exercise models to maximize athletes’ training ability as exercise media. Material and Methods: This research is experimental research with a “Two Groups Pretest-Posttest” design that includes a pre-test before the subjects are given treatment and a post-test after the treatment. The research populations were twenty male athletes from a basketball club in Yogyakarta, selected using a random sampling technique. All samples were subjected to a pre-test to determine the treatment group, ranked by their pre-test scores, then matched with the A-B-B-A pattern in two groups with ten athletes each. The sampling technique used in this dividing step was ordinal pairing. This research was conducted 18 times in treatment. The instrument used was an agility test using the Lane agility test. The movements for each number are as follows, (1) Sprint, stunt step, (2) Right slide (running with a guard position to the right side), (3) Run backwards, (4) Left slide, (5) Right slide, (6) Sprint, stunt step, pivot, (7) Right slide, (8) Sprint. Results: The research used normality, homogeneity, and hypothesis testing. To test the hypothesis, the researchers used the two types of t-tests, namely paired sample test and the independent sample test. The T-test is a statistical analysis technique that can be used to determine whether there is a significant difference between two sample means or not. The results revealed that the t count was 2,335 with a p significance value of 0.031. Because the t-count was 2.335 and the significance value was <0.05, these results indicated a significant difference. Conclusions: There was a significant difference in the effect of the exercise methods (water exercise and sand exercise) on agility. The sand exercise training method was higher (better) than the water exercise training method on the agility of basketball athletes.