The focus of this thesis is on social workers’ role in assessing and making decisions in child welfare cases. The primary aim is to identify factors that influence decisions concerning reports assessed, investigated, dismissed or processed to intervention within child welfare agencies in Swedish municipalities. Social policy, professional and organisational factors are perspectives considered in the analysis. All reports and requests for support for children and adolescents in the 0–19 age group were collected during two months in two local agencies, in 2003 (n= 260) and followed by interviews with social workers. Factors connected to social workers’ assessments at different stages in the process were tested in regression models and grounds for their assessments explored. Main results: two-thirds of reports are sorted out without investigation. One-fifth led to interventions. The highest probability for a report to be investigated was if it was assessed as acute, concerned abuse, came from a professional (not the police) concerned a girl and handled in the integrated agency. The most common problems, such as family conflicts and antisocial behaviour were investigated the least. A majority of the children came from underprivileged families, mainly poor single mothers. The process draws the pattern of a heavily tapered funnel with few interventions at the end. Children are not in focus and the attitude is to keep them out of the system for their own good. Social policy and organisational factors restrict social workers’ discretion. Contradictory demands are solved by a ‘consensual ideology’. Parallel tracks appear on risks in a narrow perspective and on voluntary counselling mainly directed to mothers. This forms child welfare into a rejecting practice, where hard social conditions are individualized. Legislation gives municipalities considerable leeway to produce a variety of services and interventions, but practice works on the basis of another kind of rationality.