This project uses experimental methods to investigate individual behavior in different contest games. A contest game is defined as a situation in which individuals or groups compete with one another by expanding effort or cost to win a specific prize. Examples range from college admissions and competition for a promotion, to global relationships in which different countries and political parties expend resources to lobby their own interests. This project consists of three parts. The first part targets the issue of asymmetric valuations between different payers and the effect of timing in contests. Contest theory predicts that more asymmetric contests generate higher competition than symmetric contests. It is also found that the effect of timing crucially depends on the order in which players make their decisions. These predictions are tested using laboratory experiments. The second part of this project studies competition between heterogeneous groups. The existing literature is built mostly on voting models in which a subjects' choice space is limited to a binary decision: whether to cast a vote or not. By contrast, in this project subjects can make their decisions on continuous strategy space. The heterogeneity between groups arises from the differences between the players within each group. The following testable hypothesis are derived from the theory: only players with the highest valuation in each group expand positive efforts in the equilibrium; contribution levels and the outcome of the contest game are independent of the number of players; and contributions do not depend on the distribution of the valuations in the group, but only on the highest valuation in each group. Finally, the third part of the project uses laboratory experiments to investigate the efficiency and potentially wasteful efforts of several alternative contests: four simultaneous and three sequential contests. The simultaneous contests include a grand contest, two multiple prize settings (equal and unequal prizes), and a contest which consist of two subcontests. The sequential contests are a sequential elimination contest, a two-stage tree contest, and a two-stage tree contest with carryovers. Alternative theoretical and behavioral models such as inequality-averse theory and quantal response equilibrium are used to explain subjects' deviations from the theoretical predictions