We analyze the effect of automation and offshorability on unemployment duration and post-unemployment outcomes such as wages and employment stability. Our rich administrative data allow us to evaluate the importance of providing unemployment training in this context. Employing a multivariate mixed proportional hazard model to deal with selectivity, we find that both the routine content in tasks as well as the probability of off-shoring negatively affects the re-employment possibilities. Labor market training is helping workers to ameliorate these negative effects and is remarkably on the spot. For workers who find re-employment, our results show that offshorability (but not automation) affects future job duration and wages positively. Our analysis reveals interesting differences by gender.