<p>We formulate and numerically solve a game-theoretic model of rational agents’ self-protective actions in an epidemic game. We prove the existence of an equilibrium and show that our model can give rise to multiple equilibria. We then compare our model simulation results with data collected from real human players in an online experiment conducted by Chen et al. (2013). Compared with game-theoretic agents, human players choose to self-protect at a higher rate and experience a lower disease prevalence. However, they receive similar endgame outcomes as measured by payoffs. In addition, human players’ decisions are dependent on their infection history, and they are less responsive to changes in disease prevalence compared to game-theoretic agents. Our results suggest that human players in the epidemic game differ substantially from fully-rational, forward-looking, strategic agents in terms of decision-making mechanisms and several measures of game outcomes.</p>