In this paper we structurally estimate a game-theoretic model of government formation in a multiparty parliamentary democracy. We focus on the timing and the terms of government agreements in the context of a multilateral stochastic model of sequential bargaining with complete information (Merlo and Wilson (1194, 1995)) where efficient delays may occur in the unique equilibrium. Besides showing that our model yields a good fit to the data on the duration of negotiations over government formation as well as government durations in postwar Italy, we use our estimates to quantify the advantage to proposing and to conduct policy experiments to evaluate the effects of changes in the bargaining procedure. We show that the gains from proposing tend to be quite large. Also, we show that changes in the proposer selection process would not affect either the duration of negotiations or government durations, while the imposition of a strict deadline would in general reduce the incentives to delay agreement as well as government durations.