Deaths of despair – alcohol-related, drug-related, and suicide mortality – have emerged as a major public health concern strongly linked to socioeconomic disadvantage. Yet much evidence relies on baseline or single-time-point measures and originates from contexts with weak social protection. Two questions remain insufficiently addressed: how work-welfare trajectories are associated with mortality risk, and how risks vary across institutionalised forms of non-employment in welfare-state contexts with strong social protection. Using linked Swedish administrative registers, we analysed 20,281 adults assessed with the Addiction Severity Index (ASI) in municipal social services during 2013-2022 – a population combining severe substance-use problems with prolonged labour-market marginality. We reconstructed annual work-welfare sequences over a ten-year window, identified trajectory clusters using sequence analysis, and estimated cause-specific mortality using Fine-Gray competing-risks models. Five trajectories emerged: employment-attached, welfare-dependent, unstable employment, health-benefit-dependent, and pension-anchored. In adjusted models, the welfare-dependent, unstable-employment, and health-benefit-dependent trajectories showed higher drug-related and non-despair mortality than the employment-attached group. Alcohol-related mortality was higher only in the welfare-dependent trajectory, while suicide risk did not differ across trajectories. Drug-related mortality gradients persisted after adjustment for baseline hospitalisation, time-varying hospitalisation, and self-reported clinical severity. These findings suggest that in a decommodifying welfare-state context, non-employment is not a uniform exposure. Institutionally structured work-welfare trajectories differentiate mortality risk, with gradients concentrated in substance-related deaths rather than suicide. The persistence of trajectory gradients after health-severity adjustment is consistent with sustained marginalisation playing a role in mortality risk beyond individual-level clinical severity, and identifies high-risk subgroups within social-service caseloads for targeted intervention.