Survivor is often credited with launching the reality television fad of the 2000s and remains a widely popular show today, yet very few empirical studies have focused on this topic. This paper examines the factors affecting whether an alliance will stay together or break apart. Our estimates show that a 10-percentage point increase in relative coalition size increases the probability that a player will stay with their coalition by 6.30 percentage points. In addition, a 10-percentage point increase in a player's voting history with players outside their coalition decreases the probability the player stays with their coalition by 6.12 percentage points. These results are robust to several alternative specifications, including both linear and non-linear models, models that separate contestants by age and gender, and models that focus on twists in the game. Further, I conclude that these results can be generalized and applied to settings outside of Survivor for example, in workplace environments where coalitions may break apart as workers progress up the employment ladder.