This paper investigates the impact of no-fault divorce laws on marriage and divorce in the United States. I propose a theory that captures the key stylized facts of the rising then declining divorce rates and the apparent convergence of divorce rates across the different divorce regimes. The empirical results suggest that a shift from fault to no-fault divorce increased the odds of divorcing for those couples who married before the shift. The analysis further suggests that those couples who marry after the shift to a no-fault regime, in turn, sort themselves better upon marriage, which offsets the direct effect of the law on divorce rates. Consistent with that selectivity argument, after a switch to a no-fault divorce regime, women get married later in life. These results hold for the law that governs property division and spousal support. The law that governs divorce grounds does not seem to matter significantly.