This study aimed to determine whether distinct subgroups of psychopathic traits exist in a sample of civil psychiatric patients, using data from the MacArthur Violence Risk Assessment Project (n = 810), by means of latent class analysis. Multinomial logistic regression was used to interpret the nature of the latent classes, or groups, by estimating the associations with criminal behaviour, violence, and gender. The best fitting latent class model was a 4-class solution: a 'high psychopathy class' (class 1; 26.4%), an 'intermediate psychopathy class' (class 2; 16.0%), a 'low affective-interpersonal and high antisocial-lifestyle psychopathy class' (class 3; 31.3%), and a 'normative class' (class 4; 26.3%). Each of the latent classes was predicted by differing external variables. Psychopathy is not a dichotomous entity, rather it falls along a skewed continuum that is best explained by four homogenous groups that are differentially related to gender, and criminal and violent behaviour.