Byline: Cigdem V. Sirin, Michael T. Koch Why are some authoritarian regimes so quick to surrender amid lower numbers of casualties while others prove willing to incur significant casualty counts to continue their war efforts? In this study, we explore the propensity of different authoritarian regime types to sustain casualties in interstate conflicts. We argue that authoritarian leaders with smaller winning coalitions find it easier to distribute the costs of militarized conflicts outside of those coalitions. This diminishes their sensitivity to casualties. Applying a theoretical model based on an inverse divide-the-dollar game (with respect to the distribution of public "bads"), we find that personalist regimes tend to sustain the highest number of casualties in militarized interstate disputes when compared to other autocracies. Our findings suggest that along with the audience cost abilities of an autocratic adversary, target states should also consider an autocratic regime's casualty sensitivity in deciding whether to reciprocate with military action. Article Note: Authors' notes: We would like to thank Scott Bennett, Brett Benson, Charles R. Boehmer, Jose D. Villalobos, and the anonymous reviewers for their instructive comments and suggestions. CAPTION(S): Table S1. Descriptive Statistics. Table S2. Sensitivity Analyses with the Minimum and Maximum MID Casualty Values. Table S3. Replicating the Analyses with Casualty Rate and Casualty Per Capita Measures. Table S4. Heckman Two-Stage Selection Models-Full Table with All Covariates. Table S5. Variance Inflation Factors. Table S6. Sensitivity Analyses with Lagged Measures. Table S7. The Number of Casualties Sustained in Militarized Interstate Disputes-Lai and Slater's Typology. Table S8. Replicating the Analyses with Alternative Casualty Codings-Lai and Slater's Typology. Table S9. Replicating the Analyses with Casualty Rate and Casualty Per Capita Measures-Lai and Slater's Typology. Table S10. Heckman Two-Stage Selection Models-Lai and Slater's Typology. Table S11. Sensitivity Analyses with Lagged Measures-Lai and Slater's Typology. Table S12. Replicating the Analyses with an Alternative Autocracy Measure "Dictator." Table S13. Sensitivity Analyses with Hybrid Regimes and Monarchies. Table S14. Sensitivity Analyses with Authoritarian Regimes' (1) Winning Coalition Size and (2) Ratio of the Winning Coalition Size to the Selectorate. Table S15. Sensitivity Analyses with Authoritarian Regimes' (1) Repression, (2) Number of Veto Players, and (3) Political Constraints. Table S16. Sensitivity Analyses with Authoritarian Regimes' (1) Competitiveness, (2) Contestation, and (3) Inclusiveness. Table S17. Sensitivity Analyses with Additional Controls. Table S18. Sensitivity Analyses with Panel Corrected Standard Errors (PSCEs).