Engel's hypothesis of pain-prone patients having a distinct pattern of developmental psychosocial experiences was tested in a controlled design including four groups of 20 patients each: a) psychogenic pain, b) organic pain, c) psychogenic bodily symptoms, d) organic disease. On admission an open-ended interview, including childhood experiences was tape-recorded. Measures were taken to minimize observer-bias. Patients in group a) had, compared to the other groups, significantly increased prevalence of: "Parents, verbally and/or physically abusive of each other", "parents, abusive of the child", "child, deflecting aggression from one parent to the other onto himself", "parents, who suffered from illnesses/-pain", "parent of the same gender as the patient suffering from pain", "pain of patient and parent in the same location", "number of operations in adulthood", "disturbance of interpersonal relationships", and "disturbance of work life". Factor analysis produced two factors explaining 73% of the variance in group a): F1 ("Brutality-Overcompensation") was related to "duration of pain", F2 ("Submission-Inhibition") to "number of operations, accidents" in adulthood.