The different phases of the leadership long cycle are hypothesized to signi
ficantly alter the number of great power rivalries that terminate and initi
ate, Specifically, the global war phase is expected to "shock" dyads into a
nd out of rivalry. Bivariate and multivariate event history techniques are
used to show that periods of capability deconcentration are associated with
increased great power rivalry terminations but not initiations. Furthermor
e, terminations are less likely to occur during phases of systemic capabili
ty concentration than in other periods, as the theory predicts. The expecta
tions concerning rivalry initiations are not supported.