OBJECTIVE:The primary purpose was to assess personality trait differences between the severely obese seeking treatment and a mainly non-obese reference group. We also investigated gender differences and differences between obese patients and obese not seeking treatment. METHOD: Personality traits were assessed using 7 of 15 scales from the Karolinska Scales of Personality (KSP): Somatic Anxiety, Muscular Tension, Psychastenia, Psychic Anxiety, Monotony Avoidance, Impulsiveness, and Irritability. Patients from the Swedish Obese Subjects (SOS) intervention study (n ¼ 3270, ages 37-57, 71% women) and the SOS reference study (n ¼ 1135, 54% women) completed the survey. Data presented in this study were gathered prior to treatment. Significance tests and effects sizes were calculated. RESULTS: Although statistically significant differences were found between obese patients and reference subjects on nearly all personality traits, effect sizes were at most moderate. Of the three scales with moderate effects sizes, differences on Somatic Anxiety and Psychastenia could be traced to items tapping condition-specific symptoms, e.g., problems with sweating and breathing as indicators of Somatic Anxiety. Moderate differences on the Impulsiveness scale (men alone) could not be explained by item composition. Further, the obese patients differed from obese in the reference group, and both obese and reference women reported significantly higher levels on Somatic Anxiety, Muscular Tension and Psychic Anxiety compared to men (effect size: small). CONCLUSIONS: Our results provided no evidence of a general obese personality profile, instead considerable heterogeneity in personality traits was observed across our obese samples (treatment seekers vs non-seekers, men vs women) and generally only small differences were noted compared to a reference study population. Further research is needed to investigate if the somewhat elevated levels of Impulsiveness, particularly among male obese patients, is affected by weight loss. When assessing personality traits in diseased groups consideration should be given to possible confounding from, e.g., somatic symptoms.