BACKGROUND: Some malignancies are known to be associated with increased risk of herpes zoster, but little is known about how associations between cancer and subsequent zoster risk vary by cancer site, by time since cancer diagnosis, and by age. METHODS: An age-, sex-, calendar time-, and practice-matched case-control study, nested in the broadly UK representative Clinical Practice Research Datalink (CPRD) primary care database, was analysed using conditional logistic regression to estimate the association between 21 of the most common specific malignancies and subsequent zoster risk. We adjusted for comorbid conditions and other potential confounders, and investigated effect modification by age and time since malignancy diagnosis. RESULTS: A total of 192 081 adult zoster patients and 732 035 controls were included. Malignancy overall was positively associated with zoster risk (adjusted OR 1.29, 95% CI 1.27-1.32), and the association was especially strong for haematological malignancies (OR 2.46, 2.33-2.60). Among specific malignancies, there was evidence that oral, oesophageal, stomach, colorectal, lung, breast, ovarian, prostate, kidney, bladder, and CNS cancers, as well as lymphoma, myeloma, and leukaemia were associated with increased zoster odds (P⩽0.05 in each case), but the magnitude of associations varied widely. The association was typically strongest within 2 years of malignancy diagnosis and decreased with older age for both haematological and solid malignancies. CONCLUSIONS: Several cancers were associated with an increased risk of zoster, particularly within the first 2 years after diagnosis and among younger individuals. Knowledge that patients with a recent diagnosis of cancer are at high risk of zoster may encourage initiation of antiviral therapy earlier in the course of zoster when the benefits are greater. Evaluation of whether patients diagnosed with cancer would benefit from early zoster vaccination is warranted.