The elevated T-maze has been developed as an animal model of anxiety to generate both conditioned and unconditioned fears in the same rat. This study explores a version of the elevated T-maze fit for mice. Inhibitory (passive) avoidance- conditioned fear-is measured by recording the latency to leave the enclosed arm during three consecutive trials. One-way escape- unconditioned fear-is measured by recording the time to withdraw from open arms. The results showed that mice do not appear to acquire inhibitory avoidance in the standard T-maze, since their latencies to leave enclosed arm did not increase along trials. Nevertheless, the open arms seemed to be aversive for mice, because the latency to leave the enclosed arm after the first trial was lower in a T-maze with the three enclosed arms than in the standard elevated T-maze. In agreement, the exposure of mice to an elevated T-maze without shield, that reduces the perception of openness, increased significantly the latencies to leave the enclosed arm. However, the absence of the shield also increased the time taken to leave the open arms when compared to that recorded in standard T-maze. Systematic observation of behavioral items in the enclosed arm has shown that risk assessment behavior decreases along trials while freezing increases. In the open arms, freezing did not appear to influence the high latency to leave this compartment, since mice spend only about 8% of their time exhibiting this behavior. These results suggest that mice acquire inhibitory avoidance of the open arms by decreasing and increasing time in risk assessment and freezing, respectively, along three consecutive trials. However, one-way escape could not be characterized. Therefore, there are important differences between mice (present results) and rats (previously reported results) in the performance of behavioral tasks in the elevated T-maze.