The object recognition task (ORT) has become increasingly popular as a memory test in neuroscience research. Scoring of ORT performance is still mostly done by hand, which can be liable to subjective scoring. To our knowledge, no suited software is available yet since the direction of the nose of the animal cannot be tracked reliably. We have developed a software paradigm that reliably tracks the nose of the rats and have conducted a series of experiments to evaluate the reliability of this newly developed program. We used Wistar rats, which showed good object memory after 1 h interval. Subsequently, we used scopolamine (SCOP) to impair the memory performance of the rats. The object exploration was scored by two observers and the automated system. Both observers and the automated system found an impairing drug effect of scopolamine on ORT performance. When using large objects the correlation between the discrimination index d2 of observers was: 0.60 (SCOP) and 0.79 (SAL). However, the correlation between observers and the automated system was quite low: 0.41 (SCOP) and 0.40 (SAL). Reducing the size of the objects increased the reliability between observers and the automated system substantially (0.82-0.87). We conclude that the use of small objects in combination with our program enables reliable automated scoring in the ORT, thus increasing the objectivity and validity of this task.