We propose an ensemble approach to evaluate mergers that combines retrospective and prospective modeling for a more reliable analysis. We begin with a retrospective analysis based on an event study of three major U.S. airline mergers and document the fragility of the findings. We then develop a structural model that nests the event study in its reduced form, clarifying the implicit assumptions in retrospective analyses that create this fragility while separating efficiency gains from changes in firms’ conduct. Using only the pre-merger data, we develop a regression-based approach that leverages exogenous changes in market structure to forecast prices after the merger. Finally, we implement structural merger analysis and show how estimates from all approaches can be synthesized for comprehensive evaluation. This methodological integration uncovers a fundamental tension: merger-induced efficiency gains were limited or, if significant, offset by increased coordination among remaining firms.