There is accumulating evidence that learning is metabolically costly. One way in which this may manifest itself is in trade-offs between learning effort and immune function, with learning increasing susceptibility to infection. We tested this idea in the context of odour learning using outbred (BKW) male laboratory mice. Mice were exposed to three experimental treatments in which they were required to learn different numbers of urinary odours. While treatment affected the extent to which mice habituated to test odours during training, differences were not a simple function of the number of odours. The fact that there was also no significant effect of treatment on the degree of preference for novel over familiar odours in subsequent tests suggests mice retained learned odour profiles equally well regardless of the number of odours. That subsequent infection with Babesia microti increased with the number of odours mice had to learn is then consistent with an increased cost to learning effort when more odours were presented. Analysis within treatments, and relationships with the change in corticosterone concentration over the period of the experiment, suggested that it was a failure to learn, rather than maintaining learning performance, in more difficult learning tasks that led to greater infection. As in a previous study of maze learning in the strain, there was no direct relationship between infection and measures of peripheral antibody (total IgG) titre. The results are discussed in relation to studies in other learning contexts and reported relationships between glucocorticoid hormones and learning outcomes.