Interim 18F-fluorodeoxyglucose positron emission tomography performed early during the course of therapy in diffuse large B-cell lymphoma is highly predictive of outcome and now used in many clinical trials to validate risk-adapted strategies. However, interpretation criteria of interim PET are not yet appropriately defined for the evaluation of tumor response and chemosensitivity. From the results of our studies, a quantitative approach based on SUV(max) reduction between baseline PET (PET0) and interim PET demonstrated a higher predictive value than visual analysis when PET was performed at two cycles (PET2) but was equivalent to visual analysis at four cycles (PET4). The SUV(max) reduction index at PET2 seems to be the best predictor of the response compared to clinical or molecular prognostic indices. When visual analysis is necessary, the use of an online independent reading network may solve the interobserver variability, but the hurdles of visual analysis deserve an international validation study to confirm the newly proposed criteria.