We examined the outcome of a cohort of patients with Hodgkin lymphoma (HL) in order to assess if fluorodeoxyglucose (FDG) positron emission tomography–computed tomography (PET/CT) at the end of treatment (end-PET) can be omitted when the interim PET (int-PET) is negative. Seventy-six ABVD(adriamycin, bleomycin, vinblastine, dacarbazine)-treated patients were retrospectively included. No change in treatment was made on the basis of int-PET results. Suspicious foci on end-PET received biopsy confirmation whenever possible. Median follow-up was 58·9 months. Uptake on int-PET higher than liver (scores 4–5) was rated positive according to the Lugano classification, while a positive end-PET corresponded to scores 3, 4 and 5. Fifteen patients had treatment failure. Sensitivity, specificity, positive predictive value (PPV), negative predictive value and accuracy of int-PET were 46·7%, 85·2%, 43·8%, 86·7% and 77·6%, respectively. For end-PET the figures were: 80%, 93·4%, 75%, 95% and 90·8%. Eight patients with negative int-PET had treatment failure; six of them were identified as non-responders with end-PET. The 5-year progression-free survival (PFS) was 87% for patients with negative int-PET versus 56% with positive int-PET. The 5-year PFS was 96% with negative end-PET versus 23% with positive end-PET. The prognostic information from int-PET as regards PFS (log-rank test P = 0·0048) was lower than that provided by end-PET (P < 0·0001). Int-PET predicted only half of the failures. When used in clinical routine, a negative int-PET study cannot obviate the need for end-PET examination.