The lateral prefrontal cortex (LPFC) is important for working memory (WM) task performance. Neuropsychological and neurophysiological studies in monkeys suggest that the lateral prefrontal cortex is functionally segregated based on the working memory domain (spatial vs. non-spatial). However, this is not supported by most human neuroimaging studies, and the discrepancy might be due to differences in methods and/or species (monkey neuropsychology/physiology vs. human neuroimaging). We used positron emission topography to examine the functional segregation of the lateral prefrontal cortex of Japanese macaques (Macaca fuscata) that showed near 100% accuracy on spatial and non-spatial working memory tasks. Compared with activity during the non-working memory control tasks, the dorsolateral prefrontal cortex (DLPFC) was more active during the non-spatial, but not during the spatial, working memory task, although a muscimol microinjection into the dorsolateral prefrontal cortex significantly impaired the performance of both working memory tasks. A direct comparison of the brain activity between the two working memory tasks revealed no differences within the lateral prefrontal cortex, whereas the premotor area was more active during the spatial working memory task. Comparing the delay-specific activity, which did not include task-associated stimulus/response-related activity, revealed more spatial working memory-related activity in the posterior parietal and premotor areas, and more non-spatial working memory-related activity in the dorsolateral prefrontal cortex and hippocampus. These results suggest that working memory in the monkey brain is segregated based on domain, not within the lateral prefrontal cortex but rather between the posterior parietal-premotor areas and the dorsolateral prefrontal-hippocampus areas.