The study investigated the working conditions associated with the accumulation of stress and lack of recovery and how recovery is related to health. The study group was employed in pre-school, home care and social work (n = 193). Recovery was assumed to be an explanatory variable for the relations between work and health. The response rate on a survey was 79 per cent. Cluster analysis identified three groups: the 'Recovered' (36 per cent of the total group) and 'Not Recovered' (25 per cent) and the 'In-between' (39 per cent). The Not Recovered displayed the whole chain of risk factors, involving difficult working conditions to which they responded with increased compensatory strategies. Despite this group having significantly greater reports of ill health, work absenteeism was not greater, which is likely related to their substituting sickness absence with sickness presence. As many as 43 per cent of the social workers were found to belong to the Not Recovered group. Multiple regression analyses controlling for background variables revealed that the Not Recovered group had a significantly higher relative risk for poor self-rated health than those in the Recovered group. Even sharper increases in relative risk existed for the other five symptoms that were analysed. Practical implications and new research questions are discussed.