Abstract In follicular lymphoma (FL), detection of bone marrow (BM) involvement (BMI) by 18F-fluorodeoxyglucose (FDG) positron emission tomography (PET)/computed tomography (CT) improves the accuracy of staging vs BM biopsy (BMB) alone. Our objective was to determine the diagnostic utility of PET for BMI FL and the prognostic value of BMI by PET (positive PET result [PET+]). Records of patients (2002-2016) with PET and BMB at the time of initial treatment were reviewed. BMI was identified by positive BMB result (BMB+) and/or unifocal or multifocal BM FDG uptake on blindly reviewed PET scans with no corresponding CT abnormality (PET+). Among 261 patients, BMI was diagnosed in 78 patients (29.9%) by PET+, in 81 patients (31.0%) by BMB+, and in 113 patients (43.3%) by either PET+ or BMB+. PET+ upstaged 24 patients to stage IV, including 10 from stages I or II to stage IV. Median duration of follow-up was 6.0 years (range, 0-16.6 years). In univariate analysis, a high Follicular Lymphoma International Prognosis Index (FLIPI) score, PET+, and BMB+ correlated with shorter progression-free survival (PFS; all P ≤ .03), and high FLIPI, PET+, and combined PET+ and BMB+ with shorter overall survival (OS; all P ≤ .01). In multivariate analysis, PET+ was the only independent predictor of PFS, whereas high FLIPI score and PET+ predicted OS (P ≤ .03). Combined PET and BMB identify BMI more accurately than either BMB or PET alone, but BMB rarely adds critical information. For patients initiating treatment of FL, identification of BMI by PET is predictive of PFS and OS.