In diffuse large B-cell lymphoma (DLBCL), the response to first-line immunochemotherapy remains somewhat unpredictable. Interim [(18)F]fluorodeoxyglucose-positron emission tomography (FDG-PET) (PET-int) analysis could be an important tool in the prompt shift to intensified regimens. We prospectively evaluated the effectiveness of PET-int carried out at mid-treatment with standard immunochemotherapy in predicting relapse in a series of 85 consecutive patients with DLBCL. PET-int results were dichotomized as positive or negative using the recently validated five-point scale scoring system. This examination was also compared with interim computed tomography (CT-int) and final PET (PET-fin). End-points were: complete remission (CR), positive predictive value (PPV) of refractoriness and relapse, negative predictive value (NPV), overall survival (OS) and progression-free survival (PFS). Observation time was fixed to 24 months unless preceded by a DLBCL-related event. The PPV of PET-int was 58% and the NPV was 77%. CR was correlated with both PET-int and CT-int (p < 0.0001), but in multivariate analysis only CT-int was correlated with CR (p = 0.002). CT-int and PET-fin were predictive of both OS and PFS, whereas PET-int was predictive only of OS (p = 0.013). In Cox regression only PET-fin was predictive for both OS (p = 0.004) and PFS (p = 0.005). PET-int was unable to discriminate those chemosensitive patients who would later relapse. We therefore believe that the use of this expensive radioactive tool is not justified as an interim analysis.