The deinstitutionalization movement is presently spreading in Europe, but studies evaluating the effects of deinstitutionalization on behaviour disturbances among people with intellectual disability (ID) are inconclusive. The prevalence of aggressive behaviour among people with ID is high in both institutions and in community. Aggression and attacks on people are a significant problem for people with ID in both institutions and society. In the present paper, the authors focus on individuals who started attacking people after deinstitutionalization. The authors studied individual and environmental characteristics before and after deinstitutionalization to look for individual and environmental predictors for the development of aggression with the hope that some could be possible intervention points for preventive action. In an institution for people with ID, all who did not attack people before deinstitutionalization were included. The individuals who started attacking others after deinstitutionalization (n = 22) were the study group (group A) and those who did not (n = 42) comprised the control group (group B). The population was examined before and after deinstitutionalization. As far as possible the same methods were used at both occasions. The covariates included individual ones, such as mental health, behaviour disturbances and behaviour deficits, and environmental ones, such as caretaker education, caretaker:patient ratio, housing and leisure activities. Psychiatric disorders were identified in 1987 and 1995 with the Psychopathology Instrument for Mentally Retarded Adults (PIMRA), which was filled in by the caretakers. Group A showed significantly more self-injurious behaviour (SIB) than group B in 1987. The sum of behaviour disturbances shown in the past year, attacks on property, SIB and other disruptive behaviours were also significantly higher in group A than in group B. The soundness scores on the PIMRA for the people in group A were lower than for the people in group B. Group A contained significantly fewer people without behavioural problems in the past year. The authors could not find any differences on other individual characteristics such as aetiologies of ID and behaviour deficits in 1987. The caretakers' evaluations of the clients need for help, based on behaviour disturbances and mental health problems, were significantly higher in group A. The people in this study differed in age, gender and degree of ID from those in most other studies on ID and challenging behaviour. The individual predictors for starting to attack others after deinstitutionalization were SIB, a slight load of other behaviour problems and a low soundness score on the PIMRA. The only environmental predictor was the caretakers' evaluation of the subjects' need for help caused by behaviour problems and mental illness. The authors could not identify possible preventive intervention points, neither individual or environmental, and therefore, further studies are needed.