The aim of this study was to determine whether it is possible to differentiate between recurrent disease and post-treatment necrosis in patients treated for a primary brain tumour. This prospective study was designed to compare the sensitivity and specificity of 201Tl single photon emission tomography (SPET) and 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography (PET) using a dual-headed coincidence camera. Sixteen patients suspected of having recurrent brain tumour (10 men, 6 women) (mean age 39.5 years, range 21-57 years) were studied. 201Tl SPET and 18F-FDG PET studies were performed on the same day. An increase in activity was considered indicative of tumour recurrence. The images were also quantified using a thallium index and an FDG index. The 18F-FDG PET images were also assessed visually using a 5-point scale. The diagnosis of tumour recurrence was based on clinical course and/or follow-up computed tomography or magnetic resonance imaging. The sensitivity of 201Tl SPET and 18F-FDG PET was 92% (11/12) and 62% (7/12) respectively. One patient initially assessed as having necrosis showed a recurrence 9 months after both studies. McNemar's analysis of these results showed a statistically significant difference (P = 0.023) in the ability of the two methods to separate with accuracy tumour from radiation necrosis. No correlation was found between the thallium index and the FDG index (r = 0.36). We conclude that 201Tl SPET is a sensitive modality for the detection of brain tumour recurrence. 18F-FDG imaging using a dual-headed coincidence camera gave significantly poorer results compared to 201Tl SPET. Our results do not justify continuation of this prospective comparative study.