BACKGROUND: Pituitary adenomas (PA) affect ~ 1:1200 of the population and can cause a wide range of symptoms due to hormone over-secretion, loss of normal pituitary gland function and/or compression of visual pathways, resulting in significantly impaired quality of life. Surgery is potentially curative if the location of the adenoma can be determined. However, standard structural (anatomical) imaging, in the form of MRI, is unable to locate all tumors, especially microadenomas (< 1 cm diameter). In such cases, functional imaging [11C-methionine PET/CT (Met-PET)] can facilitate tumor detection, although may be inconclusive when the adenoma is less metabolically active. We, therefore, explored whether subtraction imaging, comparing findings between two Met-PET scans with medical therapy-induced suppression of tumor activity in the intervening period, could increase confidence in adenoma localization. In addition, we assessed whether normalization to a reference region improved consistency of pituitary gland signal in healthy volunteers who underwent two Met-PET scans without medical suppression. RESULTS: We found that the mean percentage differences in maximum pituitary uptake between two Met-PET scans in healthy volunteers were 2.4% for (SUVr) [cerebellum], 8.8% for SUVr [pons], 5.2% for SUVr [gray matter] and 23.1% for the SUVbw [no region]. Laterality, as measured by contrast-noise ratio (CNR), indicated the correct location of the adenoma in all three image types with mean CNR values of 6.2, 8.1 and 11.1 for SUVbw, SUVbwSub and SUVrSub [cerebellum], respectively. Subtraction imaging improved CNR in 60% and 100% of patients when using images generated from SUVbw [no region] and SUVr [cerebellum] scans compared to standard clinical SUVbw imaging. CONCLUSIONS: Met-PET scans should be normalized to the cerebellum to minimize the effects of physiological variation in pituitary gland uptake of 11C-methionine, especially when comparing serial imaging. Subtraction imaging following endocrine suppression of tumor function improved lateralization of PA when compared with single time point clinical Met-PET but, importantly, only if the images were normalized to the cerebellum prior to subtraction.