Purpose: Social welfare work contains elements that may be difficult for employees to put out of their minds when the working day ends, which may affect the recovery. The purpose of this paper is to analyze the length of recovery in relation to different work characteristics and to two types of welfare work.

Design/methodology/approach: All 1,365 employees, excluding managers, of two municipality administrations were invited to a survey study. Of these, 673 (49 percent) responded. After adjusting for partial missing, the effective sample included 580 employees (43 percent). Retrospective ratings of four recovery windows were analyzed: recovery after one night’s sleep, weekends, shorter holidays and vacations.

Findings: Employees with a university education were less recovered than those with a shorter education. For those with a university education, the long arm of the job mainly involved failures regarding qualitative job demands (task difficulty). For those with a shorter education, quantitative job demands (too much to do) were most prominent for their prolonged recovery. Feedback from managers had consistent and positive associations with all four recovery windows among employees with a university education, but not among those with a shorter education for whom instead having too much to do and social support had significant spillover effects.

Originality/value: The identified differences may relate to employees with a university education having more problem-solving tasks, which may result in a higher need of work-related feedback but also in difficulties detaching from work. Thus, education and job characteristics have differential associations with self-rated recovery.