BACKGROUND: Poly(ADP-ribose) polymerase inhibitors (PARPi) maintenance therapies are used to treat advanced ovarian cancer in first line and recurrent settings. Because of concerns about associations between PARPi therapy and secondary cancers myelodysplastic syndrome (MDS) and acute myeloid leukaemia (AML), a meta-analysis of clinical trials was conducted, reporting MDS/AML incidence of 0.73 %; however, clinical trial populations are highly selective and may not reflect incidence in the wider population. METHODS: This retrospective cohort study calculated incidence of MDS/AML within five years of completing first-line chemotherapy + /- PARPi maintenance for recurrent, advanced-stage ovarian cancer. Absolute and relative risks were calculated and compared to meta-analysis. RESULTS: Of 11,531 included patients, 1529 received PARPi and 10,002 chemotherapy only. Absolute risk of MDS/AML was 0.3 % (n = 5/1529) for chemotherapy + PARPi maintenance therapy versus 0.1 % (n = 10/10,002) for chemotherapy alone. Relative risk was 2.97 (95 % CI 1.02, 8.68, p = 0.046) in patients receiving PARPi maintenance versus chemotherapy alone. DISCUSSION: Relative risk of MDS/AML was greater in patients treated with PARPi; however, absolute risk was low in both treatment groups and lower than in the meta-analysis of trials. This analysis suggests small increased relative risk of MDS/AML associated with PARPi maintenance versus chemotherapy only, but not increased absolute risk.