Objectives. We tested 2 hypotheses found in studies of the relationship between suicide and unemployment: causal (stress and adversity) and selective interpretation (previous poor health).

Methods. We estimated Cox models for adults (n = 3 424 550) born between 1931 and 1965. We examined mortality during the recession (1993–1996), postrecession (1997–2002), and a combined follow-up. Models controlled for previous medical problems, and social, family, and employer characteristics.

Results. During the recession there was no excess hazard of mortality from suicide or events of undetermined intent. Postrecession, there was an excess hazard of suicide mortality for unemployed men but not unemployed women. However, for unemployed women with no health-problem history there was a modest hazard of suicide. Finally, there was elevated mortality from events of undetermined intent for unemployed men and women postrecession.

Conclusions. A small part of the relationship may be related to health selection, more so during the recession. However, postrecessionary period findings suggest that much of the association could be causal. A narrow focus on suicide mortality may understate the mortality effects of unemployment in Sweden.