Alterations in hippocampal cholinergic activity associated with different types and/or stages of learning were explored using measures of sodium-dependent high-affinity choline uptake (SDHACU) in the hippocampus of C57BL/6 mice. Animals were divided into 'active' subjects submitted to memory testing before being killed and 'quite' controls. 'Active' subjects were trained in a radial-arm maze on either a discrimination task (mixed Working Memory (WM)-Reference Memory (RM) task) or a Delayed-Non-Matching-To-Place task (more selective WM-task). In the discrimination task mice were sacrificed after either the 1st, 2nd, 3rd or 9th daily session of training and at intervals of either 30 s, 15 min, 24 h or 9 days post-test. In the DNMTP-task all subjects of the 'active' groups were sacrificed after 12 days of training at either 30 s or 24 h post-test. Results showed that: (1) Both types of training induced an immediate (30 s post-test) increase of hippocampal SDHACU as compared to 'quite' control condition. (2) In the discrimination task, this immediate increase in SDHACU was followed by a decrease leading to a long-lasting (24 h and 9 days) inhibition of this cholinergic marker. This secondary decrease in SDHACU occurred earlier (15 min post-test) at the end (9th session) than at the beginning (1st-3rd sessions) of training. Thus, as training progressed there was a shortening of the testing-induced cholinergic activation. (3) By contrast, in the DNMTP-task, SDHACU was still increased at the interval of 24 h following the last session of DNMTP-training. (4) The amplitudes of both the immediate (30 s) increase and subsequent secondary (15 min) decrease in SDHACU after the last (9th) session of discrimination training were significantly related to the rate of acquisition and behavioural profile for individual animals. Subjects that had displayed better response accuracy across the 9 days of training exhibited the highest SDHACU at 30 s post-test and the lowest at 15 min post-test. These results are discussed in the context both of previous findings on the effects of training on cholinergic activity, and of contemporary models of hippocampal function. It is suggested that (1) an increase in hippocampal cholinergic transmission during testing would facilitate the acquisition of a 'relational' kind of informations (spatial WM and RM); (2) the post-training consolidation (spatial RM) of information would be facilitated by a decrease and long-lasting inhibition of hippocampal cholinergic activity.