In a variety of taxa, individuals behave in consistently different ways. However, there are relatively few studies that empirically test the potential mechanisms underlying the causes and maintenance of these personality differences. Several hypotheses for the causes and maintenance of risky personality traits have been suggested but all have received mixed support. Both the pace-of-life hypothesis and state-dependent safety hypothesis propose that differences in internal state cause and maintain personality traits. Formally, the pace-of-life hypothesis states that differences in life-history traits including productivity (growth) and residual reproductive value (age) create initial differences in individual behaviour that is later maintained by positive feedback, while the state-dependent safety hypothesis suggests that body condition (mass) is responsible for causing and maintaining behavioural differences. We tested and evaluated whether either of these two hypotheses explained the causes or maintenance of variation in risk-related personality traits –defensive aggression, activity and exploration– in yellow-bellied marmots, Marmota flaviventer. We found little support overall for these hypotheses in explaining maintenance in activity or exploration. However, for defensive aggression, we found positive feedback for both mass and age.