Between 2018 and 2022, Nigeria experienced continuous transmission of circulating vaccine-derived type 2 poliovirus (cVDPV2), with 526 cases of cVDPV2 poliomyelitis detected in total and approximately 180 million doses of monovalent type 2 oral poliovirus vaccine (mOPV2) and 450 million doses of novel type 2 oral poliovirus vaccine (nOPV2) delivered in outbreak response campaigns. Inactivated poliovirus vaccine (IPV) was introduced into routine immunisation in 2015, with a second dose added in 2021. We aimed to estimate the effectiveness of nOPV2 against cVDPV2 paralysis and compare nOPV2 effectiveness with that of mOPV2 and IPV. In this retrospective case-control study, we used acute flaccid paralysis (AFP) surveillance data in Nigeria from Jan 1, 2017, to Dec 31, 2022, using age-matched, onset-matched, and location-matched cVDPV2-negative AFP cases as test-negative controls. We also did a parallel prospective study from March, 2021, using age-matched community controls from the same settlement as the cases. We included children born after May, 2016, younger than 60 months, for whom polio immunisation history (doses of OPV from campaigns and IPV) was reported. We estimated the per-dose effectiveness of nOPV2 against cVDPV2 paralysis using conditional logistic regression and compared nOPV2 effectiveness with that of mOPV2 and IPV. In the retrospective case-control study, we identified 509 cVDPV2 poliomyelitis cases in Nigeria with case verification and paralysis onset between Jan 1, 2017, and Dec 31, 2022. Of these, 82 children were excluded for not meeting inclusion criteria, and 363 (85%) of 427 eligible cases were matched to 1303 test-negative controls. Cases reported fewer OPV and IPV doses than test-negative controls (mean number of OPV doses 5·9 [SD 4·2] in cases vs 6·7 [4·3] in controls; one or more IPV doses reported in 95 [26%] of 363 cases vs 513 [39%] of 1303 controls). We found low per-dose effectiveness of nOPV2 (12%, 95% CI -2 to 25) and mOPV2 (17%, 3 to 29), but no significant difference between the two vaccines (p=0·67). The estimated effectiveness of one IPV dose was 43% (23 to 58). In the prospective study, 181 (46%) of 392 eligible cases were matched to 1557 community controls. Using community controls, we found a high effectiveness of IPV (89%, 95% CI 83 to 93, for one dose), a low per-dose effectiveness of nOPV2 (-23%, -45 to -5) and mOPV2 (1%, -23 to 20), and no significant difference between the per-dose effectiveness of nOPV2 and mOPV2 (p=0·12). We found no significant difference in estimated effectiveness of the two oral vaccines, supporting the recommendation that the more genetically stable nOPV2 should be preferred in cVDPV2 outbreak response. Our findings highlight the role of IPV and the necessity of strengthening routine immunisation, the primary route through which IPV is delivered. Bill & Melinda Gates Foundation and UK Medical Research Council.