The statistical relationship between accepted job duration and the job-to-job transition strategy chosen by job quitters is analyzed using parametric duration methods. Because job quitters can sort themselves between unemployed and employed search, the labor-market state occupied before acceptance of a new job must contain information about the search-preferred strategy. The results show that jobs preceded by unemployment tend to be shorter but also reveal a substantial amount of heterogeneity when the sample is split between those having comparative advantages in unemployed search and those with comparative advantages in employed search. However, even for those who have comparative advantages in unemployed search, choosing unemployment does not seem to raise subsequent job duration.