Can an eye for an eye turn the whole world sanctioned?

International sanctions have been an increasingly common tool for enforcing international norms of behavior, among other goals. There is continuing debate on what drives sanctioning behavior: while IR theories of sanctions have a long empirical history, more recent studies identify a need to extend these theories to better account for the endogenous nature of state sanction networks. Using a combination of network and IR-based insights, we aim to build a theory-driven, interpretable model of international sanctions that has high predictive utility. Using a separable version of dynamic network logistic regression, we test network theories of “Matthew effects”, reciprocity, and previous state-level sanctioning activity alongside traditional IR theories regarding the democratic peace, cultural or institutional similarity, power imbalance, and trade. Though we find that mechanisms from established IR theories largely hold with the inclusion of network endogeneity, endogenous network effects are more powerful than traditional IR concepts for predictive accuracy of the sanctioning network. We also find considerable differences for factors driving the imposition of sanctions (tie formation) versus persistence of sanction regimes (persistence), pointing to the importance of treating such effects separately.