Popular perception holds that employment stability declined towards the end of the twentieth century. However, most studies conclude that the proportion of long term jobs has remained remarkably stable over the last few decades. This study focuses on this discrepancy by tracking self-reported changes in Canadian employment durations over an extended period. This is done in order to reconcile popular perception with recent studies and nest the existing literature in a broader historical context. The study makes use of finite mixture decompositions on successive cohorts of employees starting from the 1950s to identify worker types within cohort-based distributions. Then, using tests of stochastic dominance, it is shown that the distribution of employment has indeed changed. Furthermore, detailed examination by employment spell and birth cohort is used to identify contributing factors to the identified declines in stability. It is surmised that structural changes in the economy and broader society played a large role in the seeming reduction in employment stability. While for men these shifts were clear, for women the evidence was more mixed.