The choice of a rituximab-based regimen and the prognostic significance of interim 2-[fluorine-18]fluoro-2-deoxy-D-glucose-positron emission tomography/computed tomography (FDG-PET/CT) in primary mediastinal large B cell lymphoma (PMBCL) are debatable. We evaluated the clinical features and outcomes of 95 consecutive patients with PMBCL who were treated between 1985 and 2009. Forty-three patients received rituximab-based chemotherapy, R-VACOP-B (N = 30) or R-CHOP21 (N = 13), whereas 52 patients were treated with VACOP-B (N = 47) or CHOP21 (N = 5). Radiotherapy was not given. Patients who received rituximab had a 5-year progression-free survival (PFS) of 79 % and overall survival (OS) of 97 % compared with 58 % (p = 0.06) and 88 % (p = 0.2), respectively, without rituximab. Five-year PFS in patients treated with R-VACOP-B, R-CHOP21, VACOP-B, and CHOP21 were 83, 69, 62, and 20 %, respectively (p = 0.039). However, direct comparison showed that the difference between PFS rates in patients receiving R-VACOP-B compared to R-CHOP21 was not statistically significant (p = 0.3). None of the standard clinical risk factors predicted for PFS and OS in patients receiving rituximab (R)-chemotherapy. Mid-interim FDG-PET/CT scans were performed in 30/43 patients who received R-chemotherapy. The negative predictive values of mid-PET activity were high (100 % for R-VACOP-B and 86 % for R-CHOP21) while the positive predictive values (PPV) were relatively low (30 and 75 %, respectively). Despite the low PPV, the 5-year PFS for mid-PET-negative patients (N = 16) was significantly higher (94 %) than that for mid-PET-positive (N = 14) patients (57 %, p = 0.015). This retrospective analysis demonstrates that the superiority of VACOP-B over CHOP21 for treatment of PMBCL disappeared once rituximab was added. The potential benefit of using interim PET activity as a guide for continuing therapy in patients with PMBCL remains unclear due to the relatively low PPV.