The advantages of positron imaging in diagnosis of various malignancies have been touted by proponents of positron emission tomography (PET) technology. While PET lacks the anatomic resolution of CT and MR imaging, it is quite sensitive to small foci of increased metabolic activity, such as small tumors.ECRI conducted a short-form technology assessment of PET for diagnosis of recurrent brain tumor, comparing it to single photon emission CT (SPECT) and to CT and MRI.A total of 22 published journal articles with PET clinical trial results were examined. Eight of these met our criteria for analysis:$#183; Study was reported in the English-language peer-reviewed literature, as a full article rather than an abstract only$#183; Study compared results of the positron imaging modality to results of a reference test (surgical pathology or clinical follow-up) in patients with recurrent brain tumor and in patients with radionecrosis or other treatment-related changes.$#183; Interpretation of the positron results was blinded to the reference standard$#183; Study reported sufficient data to permit calculation of sensitivity and specificity.$#183; Study did not include any pre-treatment patients, or reported them separately.$#183; At least ten patients total in recurrent tumor and control groups.Studies were included irrespective of their prospective or retrospective design, and irrespective of the role of MR or CT in acquiring or analyzing PET results.Meta-analysis of these trials yielded a summary ROC with a very broad confidence interval. The summary ROC derived from five clinical trials of thallium-201 SPECT also had a very broad confidence interval.14 of the 22 PET studies reported that CT and/or MRI was used in a complementary role to PET: CT and MR images were used to identify lesions to be analyzed with PET. 10 of the 17 SPECT trials also used CT and/or MRI for localization purposes.We concluded that PET appears to have some capability for differential diagnosis between brain tumor recurrence, but there is insufficient evidence for proving this hypothesis or for estimating the sensitivity and specificity of PET. We were unable to draw any conclusions about whether PET is more effective than SPECT for this particular diagnostic task.