BACKGROUND: Poor self-control is a strong correlate of criminal propensity. It is conceptualized and operationalized differently in criminology than in other scientific traditions.

AIMS: (1) To verify the dimensionality of the criminological Grasmick self-control items, other self-regulation items and morality ones. (2) To re-interpret the dimensions using a clinical perspective, a taxonomic/diagnostic model and references to possible "biological underpinnings." (3) Validate the dimensions by associations with crime.

METHOD: Population: all persons born 1995 in Malmö and living there at age 12. A random sample (N = 525) filled in a comprehensive self-report questionnaire on themes like personality, crime/abuse and social aspects at age 15, 16 and 18. Age 18 data were analysed: 191 men and 220 women.

RESULTS: Self-regulation items were 4-dimensional: ADHD problems (Behavior control and Executive skills) and two Aggression factors. Morality items formed a fifth dimension. Negative Affect and Social interaction factors covered the rest of the variance. The validity of these factors was backed up by correlations with similar items/factors. Self-regulation subscales predicted crimes better than the Grasmick scale; an interaction with morality improved prediction still further. Sex differences were over-all small with three exceptions: Aggression, Morality and Negative affect.

CONCLUSION: We identified four dimensions of the 20-item Grasmick instrument: Cognitive action control (impulsiveness/sensation seeking, response inhibition), Executive skills/future orientation, Affective/aggression reactivity and Aggression control. All should be possible to link to brain functional modules. Much can be gained if we are able to formulate an integrated model of self-regulation including distinct brain functional modules, process-and trait-oriented models, relevant diagnoses and clinical experiences of individual cases.