This paper analyzes the behavior of firms with defaulted credits in terms of recovery or extinction. By defining classes for the severity of default, survival models for the multiple transitions from each class are estimated. The models are used to simulate the evolution of a firm’s credit conditional on its characteristics. Estimates for the expected recovery or extinction rates are constructed from these simulations. They show that (i) the severity of default strongly influences the probability of extinction; (ii) for less severe default episodes, recovery is faster than extinction, and the opposite is true for more severe defaults; (iii) larger firms tend to display better outcomes; (iv) and the number of employees is the single most important determinant of the time profile of the extinction/recovery process. Estimates of a loss given default measure suggest that the supervision recommendations found in the literature are appropriate.