Gullies experience varying states of activity during their life cycle. For example, their highest growth rates are commonly observed in the period that follows their initiation, whereas they are less active when reaching stability.Understanding the environmental conditions under which gullies initiate, expand, and stabilize is therefore vital to mitigate their impacts. Data-driven susceptibility assessments are key approaches to understanding these conditions at the catchment scale. However, such assessments commonly focus, at best, only on one part of the problem (e.g., on the gully heads) and do not cast a critical eye on the significance of the processes they would highlight. So far, no study has attempted to explicitly model the life cycle of gullies. Here we address this research gap through modelling separately the location where new gullies initiate and where they stabilize using both gully initiation points and gully heads. More specifically, we study over 4400 active and inactive gullies in the Southern Main Ethiopian Rift. Using logistic regression models, we assess the susceptibility to gully initiation points derived from slope-drainage area (S-A) thresholds. This is then compared with the susceptibility of active or inactive gully heads at the level of four catchments considered together and separately. Highly susceptible areas for gully initiation are mainly located in rejuvenated landscapes downslope of rifting-associated knickpoints, where steep hillslopes are more recent than those of the surrounding relict landscapes. Concave slopes with a higher surface runoff concentration favour initiation of gullies. In contrast, gullies stabilize in convex slopes with a more diffusive characteristic. The resulting susceptibility models are a step towards contributing to the decisionmaking process on the optimized locations of soil and water conservation measures during several stages of the life cycle of gullies.