BACKGROUND: Spatial memory and navigation have been widely studied using animal models, most of them were performed with rodents and only few of them with non-human primates. Also, most of these studies were achieved in simple and small-scale mazes. NEW METHOD: In order to test navigation strategy in macaque monkeys, we used a double-cross virtual maze in which they could freely move using a joystick and a procedure adapted from previous studies with rodents. The monkeys had first to learn several routes from a specific starting point to reach separate positions, each one containing distinct symbols, and then to associate each route with these symbols. After this phase, the monkeys had to reach these targets from a second starting point positioned at the other end of the maze, requiring the use an allocentric strategy. In order to evaluate monkey's strategy, we analyzed monkey's first choice as a function of the symbol presented. RESULTS: The monkeys were able to learn routes and to associate routes with symbols but were not able to use an allocentric strategy. They also did not use the procedural strategy and seemed disoriented or adopted a stereotypical behavior. COMPARISON WITH EXISTING METHOD(S): Contrary to the few studies using small-scale environments, the monkeys seemed unable to solve the allocentric task. The studies using virtual mazes seemed not to specifically analyze monkey's navigation strategy. CONCLUSIONS: Supplementary experiments will be necessary to explain the origin of the monkey's inability to use an allocentric strategy. Modified learning conditions could also incite the monkeys to develop an allocentric representation.< Réduire