In this paper, we study the initial-conditions problem, a complication associated with left-censored or interrupted spells in the econometric analysis of labor market transitions. In the presence of unobserved individual-specific heterogeneity, no consistent estimators have been previously constructed. This paper proposes such an estimator using indirect inference (II). The II procedure simulates the structural model and “matches” the simulated data with the actual data via the implementation of an informative auxiliary model. Consistency and asymptotic normality of the II estimator are proved. Monte Carlo experiments as well as a real data set are used to illustrate the small-sample performance of the II estimator. These results show that the II estimator is insensitive to the alternative auxiliary models chosen for the II estimation.