BACKGROUND Although athletes with mental retardation (MR) have achieved an important role, literature concerning the evaluation tests in basketball is still poor. OBJECTIVE To assess basketball ability before and after a 4-month training performed before championship for athletes with MR and to correlate ability variations with MR levels. METHODS 15 trained basketball players with MR (11 men and 4 women; age range 21-43 years; MR: 3 mild, 8 moderate, 8 severe and 1 profound). Athletes were tested before (pre) and after (post) 4 month training preceding the championship. The tests assessed 4 levels of ability, each one characterised by 4 fundamental areas of this game: ball handling, reception, passing and shooting. Each area was divided into 5 specific components. RESULTS The team average score, based on the score of each athlete's 4 levels, improved by 6.6 points (41.6 (SD 11.9) vs 48.2 (14.7)). The comparison between pre and post scores in each level showed increases (p<0.01), especially in level II (14.4 (3.5) vs 16.5 (3.3)) and in level III (7.1 (5.7) vs 9.2 (6.6)). Within level II, ball handling (3.67 (1.2) vs 4.37 (0.5); p<0.05) and passing (3.20 (1.2) vs 3.97 (1.3); p<0.01) improved; in level III reception (3.21 (1.1) vs 3.73 (1.1); p = 0.01) and shooting (1.82 (1.1) vs 2.45 (1.3); p<0.05) increased. CONCLUSIONS A 4 month training caused a general improvement, especially evident in levels II and III. Total score reached in level II was negatively correlated with MR level (r = -0.56; p<0.05), indicating that athletes with lower MR obtained higher scores.