In fencing, it remains unclear whether practice enhances visual‐spatial attention allocation. We explored whether this ability is improved in fencers and if it is related to attack speed and accuracy. Twelve novices (<1 year of experience) and 12 trained epee fencers (18 ± 10 years of experience) visited the laboratory twice (familiarization and testing session). They performed a covert orienting of visuospatial attention test (COVAT) on a computer and an epee test, involving 30 trials of 3 shuttles followed by fast attack phases where the fencers quickly hit a target (randomly proposed out of 8). We measured COVAT reaction time, number of successful target hits, and execution time to hit in the fencing test. We found shorter COVAT reaction time for trained fencers (332 ± 24 ms) versus novices (367 ± 32 ms; p < 0.001). The number of hits was greater for trained fencers (22 ± 3) versus novices (16 ± 3; p < 0.001). ANCOVA showed a difference in execution time at the test (823 ± 73 ms vs. 913 ± 141 ms, p = 0.035). A relationship was found between hits and execution time and between execution time and COVAT reaction time for the trained group ( r = 0.62, p = 0.03 and r = 0.70, p = 0.01, respectively) but not in the novice group ( r = 0.11, p = 0.72 and r = 0.45, p = 0.14, respectively). Mediation analysis showed that the relationship between execution time and number of hits (ADE: p = 0.008) was not mediated by COVAT reaction time (ACME: p = 0.17). These results evidence the importance of visual‐spatial attention allocation in fencing and evidence differences between novices and trained fencers with important implications for talent development in the early career stage.