The Narcissistic Personality Inventory (NPI) is commonly used in empirical studies of narcissism. Few population-based studies have been published. Our aim was to do a confirmatory factor analysis (CFA) of the NPI 29 item version with a four-factor structure, in two population-based samples and in a patient sample, and present normative population-based data. The NPI-29 was filled in by 324 respondents from the Norwegian population, 231 from the Swedish population and 167 Norwegian psychiatric patients. The four-factor structure of the NPI-29 with Leadership/Power, Exhibitionism/Self-admiration, Superiority/Arrogance and Uniqueness/Entitlement was reproduced in these samples. The CFA models showed good fit indices in all samples. Mean scores on the NPI-29 and four subscales hardly differed between the samples. For the NPI-29 total score and factors, few significant differences were observed. CFA of the samples supported the factor structure of the NPI-29 formerly identified by principal component analysis of the Swedish population sample.