The tax regimes applied to couples in many countries including the US, France, and Germany imply either a marriage penalty or a marriage bonus. We study how they affect the decision to get married by considering two potential spouses who play a marriage proposal game. At the end of the game they may get married, live together without formal marriage, or split up. In this signaling game, proposing (or getting married) is costly but can indicate strong love. The striking property we obtain is that a marriage bonus may actually reduce the probability that a couple gets married. If the bonus is sufficiently large, the signaling mechanism breaks down, and only a pooling equilibrium in which fewer couples get married remains. Similarly, a marriage penalty may increase the marriage probability. Specifically, the penalty may lead to a separating equilibrium with efficiency enhancing information transmission, which was otherwise not possible. Our results also imply that marriage decisions in the laissez-faire are not necessarily privately optimal. In some cases a bonus or a penalty may effectively make the marriage decision more efficient; it may increase the number of efficient marriages that otherwise may not be concluded.