An olfactory span task that required rats to discriminate an olfactory stimulus added to an increasing list of such stimuli (nonmatching-to-sample; NMTS) was employed to assess the role of the basal forebrain cholinergic system in the animals' olfactory working memory capacity. A separate group of animals was trained in a matching-to-sample (MTS) version of this task that did not tax span performance. NMTS animals required significantly more sessions to reach an olfactory span of 18 stimuli than MTS rats. Infusions of the cholino-immunotoxin 192 IgG-saporin into the basal forebrain resulted in decreases of cortical acetylcholinesterase (AChE)-positive fibre density ranging from 80% in frontodorsal and frontoparietal regions to 35% in the pyriform cortex and 24% in the olfactory bulb. Postsurgery span performance was significantly reduced in lesioned NMTS but not MTS animals. Span performance in lesioned NMTS animals recovered following 4 weeks of postoperative training; however, these animals' span remained vulnerable to the effects of increased intertrial intervals. The distribution of errors in lesioned animals indicated a recency effect. In NMTS animals, olfactory span performance during the initial two postoperative weeks correlated significantly with AChE-positive fibre density in neocortical but not olfactory areas. The privileged, automatic processing of olfactory stimuli in rats may have contributed to the transience of the lesion effect. The results support the crucial role of cortical cholinergic input in the mediation of aspects of processing capacity.