The present study was designed to examine which kind of memory: reference or working, better correlates with individual variation in rats' spatial learning abilities. To answer this question two groups of rats were trained to an arbitrary criterion in a partially baited 12-arm radial maze under two different experimental conditions: with or without allothetic cues. After 10 days break, rats were examined under the same conditions for memory retention. Within- and between-group variation in the length of training to criterion, and in the frequency of reference and working memory errors were analysed. The present experiment confirmed the facilitating effect of the presence of distal visual cues on place learning in rats. Task-dependent (between-group) differences in the rate of learning were attributed to differences in the frequency of reference memory errors. Conversely, within-group variation in the rate of task acquisition reflected individual variation in the frequency of working memory errors. These results were looked upon from an evolutionary perspective. Low correlation between reference and working memory errors confirms that these two types of memory have different mechanisms. The fact that differences in the rate of learning were not paralleled by the differences in the memory retention supports the notion that memory acquisition and memory retention are two independent processes.