Assessments of historical contingency have advanced our understanding of adaptive radiation and community ecology, but little attention has been given to assessing the importance of historical contingency in population ecology. An obstacle has been the unmet need to conceptualize historical contingencies for populations in a manner that allows for their explanatory power to be assessed and quantified so that it can be directly compared with the explanatory power of statistical models representing other hypotheses or theory-based explanations. Here we conceptualize historical contingencies as a series of random events characterized by (1) significant legacy effects that are comparable in length to the waiting time between such events, and (2) the disparate nature of individual events in the series. From that conceptualization, we present a simple quantitative framework for assessing the explanatory power of historical contingencies in population ecology and apply it to an existing long-term dataset on the predator-prey system in Isle Royale National Park. The population-level phenomenon that we focused on was predation rate because it is a synthesis of three basic elements in population ecology (predator abundance, prey abundance and kill rate). We also compared the explanatory power of models of the historical contingency hypothesis to a wide-range of alternative, theory-based, statistical models used to assess underlying mechanisms or forecast future dynamics. Models of the historical contingency hypothesis explained over half of the interannual variation in predation rate and performed similarly, or better than, the vast majority of alternative, theory-based, models. Those findings highlight the potential value of reconsidering the way that population ecologists traditionally attempt to explain phenomena. We also discuss how this new conceptualization of the historical contingency hypothesis can also be valuable for synthesizing several other important ecological concepts of broad significance, especially reddened spectra, tipping points, alternative stable states, and ecological surprises. If the historical contingency hypothesis were found to be broadly applicable, then it would likely explain why ecologists are conspicuously poor at forecasting future dynamics.