Multiple theories posit the existence of a dyadic democratic peace. The authors extend the logic of three theories of the democratic peace ĝ€" informational, normative, and preferences ĝ€" and find that they make different predictions with respect to the onset and escalation of disputes across the range of similar regime dyads. First, regarding dispute onset, the preferences argument, but not the normative and informational arguments, expects autocratic dyads of similar type to have less conflict onset than mixed dyads. Second, the normative argument expects democratic, but not non-democratic, dyads to be less likely to escalate their disputes, while the informational argument expects democracy to have little impact, after conflict onset has been taken into account. The preferences argument expects all dyads of similar regime type to be less likely to escalate their disputes. Critical tests of these expectations are conducted by estimating a censored choice model of conflict onset and escalation, using multiple measures of interstate conflict. The authors find little support for a broader regime-similarity peace, and their findings on democratic dispute escalation favor the informational argument over the normative argument.