PurposeIn the current study, the authors explore changes in multiple police officer-involved deaths (MOIDs) and on changes in the racial makeup of MOID victims in different stages of implementation of a duty-to-intervene (DTI) policy by the New York City Police Department (NYPD).Design/methodology/approachThe authors use fatal encounters to analyze data on MOIDs involving NYPD officers from 2000 to 2019, including three time periods: pre-DTI, initial DTI, and revised DTI. The authors use non-equivalent dependent variables design and t-tests to determine the significance of differences in MOID occurrence and civilian race between these periods.FindingsThe greatest reduction in MOIDs was observed during the initial DTI period, followed by an uptick in MOIDs during the revised DTI period. We also observed that MOIDs are racialized events that disproportionately impact Black New Yorkers. This remained the case even after the implementation of DTI.Research limitations/implicationsThe authors find mixed support for DTI as an administrative control for preventing MOIDs and reducing racial disparities in MOIDs. DTI implementation period, the significant reductions in MOIDs in the initial DTI period, but not the second also lends support for the notion that community pressure (i.e. resurgence of Black Lives Matter) also impacts officer behavior. This suggests that along with strong administrative controls, the behavior of the public may also be an important factor in controlling officer behavior.Originality/valueThis article contributes to the growing literature on duty to intervene and offers an early investigation into DTI as an administrative control for MOIDs using the NYPD as a case study. The authors examine changes in MOIDs and the racial makeup of civilians who were killed in MOIDs in three DTI periods. To the authors’ knowledge, no other study has done this.