Prosocial behaviour includes a wide range of actions such as helping, sharing, comforting and cooperating. These actions are characterized by a concern for the rights, feelings and welfare of other people. The term originated during the 1970s as an antonym of antisocial behaviour. Prosocial behaviour of school children aged between 6 and 16 years is a complex phenomenon resulting from factors, including biological, psychological and social factors. The main objective of this paper is to identify home-, class-, school- and individual-related factors that are significantly related to prosocial behaviour and develop new ways of understanding, promoting and enhancing this behaviour in Maltese schools. The Strengths and Difficulties Questionnaire (Goodman, Journal of Child Psychology and Psychiatry, 38, 581–586, 1997) was used to generate the prosocial behaviour scores. A proper methodology for analyzing hierarchical structured data where observations are nested within groups is multilevel modelling. This paper presents a three-level random intercept models that accommodate random effect within each level of nesting and examines the contribution of a number of predictors in explaining variations in the prosocial behaviour scores elicited from 1578 students attending primary and secondary state, church and independent schools. The model identifies individual-related variables, particularly engagement, diagnosis and intervention, as better predictors of prosocial behaviour than home-, classroom- and school-related variables.