Although histological diagnosis is indispensable in treating primary central nervous system lymphoma (PCNSL), we sometimes face an intractable situation in which tissue can be obtained only from a deep-seated brain lesion. In place of a histological diagnosis, the diagnostic adequacy of the combined use of 18 F-FDG PET and corticosteroid administration for PCNSL located in a deep-seated brain structure is reported. Patients with a deep-seated tumor were treated as having PCNSL without histological confirmation, based on the following criteria: (1) there was no evidence of systemic malignancy; (2) the tumor showed an extremely high FDG uptake relative to normal gray matter on pretreatment 18 F-FDG PET; (3) the tumor decreased in size 1 week after diagnostic therapy by corticosteroid administration on contrast-enhanced T1-weighted magnetic resonance imaging (MRI). FDG uptake of the lesion was evaluated by the maximum of standardized uptake values (SUVmax) and tumor-to-normal ratio of the SUV (T/N ratio). The extent of the tumor reduction was calculated by volumetric analysis for the treatment response to corticosteroid administration. Ten patients (4 males and 6 females) matched these criteria. On pretreatment 18 F-FDG PET, mean SUVmax in the tumor was 24.8 (8.75-60.75), and mean T/N ratio was 3.24 (2.17-5.12). The extent of tumor volume reduction was shown to be 21 to 68 % 1 week after diagnostic therapy by corticosteroids. Mean total dose and duration of corticosteroids were 719 mg as prednisolone and 6.5 days, respectively. Nine patients achieved complete response and one patient achieved partial response on MRI after standard treatment for PCNSL with high-dose methotrexate and/or whole-brain irradiation. Although the value of biopsy is universal, combining 18 F-FDG PET and corticosteroid administration is an important alternative method that may lead to the diagnosis of deep-seated PCNSLs in cases with intractable histopathological confirmations.