Background Paranoid ideation has been regarded as a cognitive and a social process used as a defence against perceived threats. According to this perspective, paranoid ideation can be understood as a process extending across the normal–pathological continuum. Methods In order to refine the construct of paranoid ideation and to validate a measure of paranoia, 906 Portuguese participants from the general population and 91 patients were administered the General Paranoia Scale (GPS), and two conceptual models (one‐ and tridimensional) were compared through confirmatory factor analysis (CFA). Results Results from the CFA of the GPS confirmed a different model than the one‐dimensional model proposed by Fenigstein and Vanable, which comprised three dimensions (mistrust thoughts, persecutory ideas, and self‐deprecation). This alternative model presented a better fit and increased sensitivity when compared with the one‐dimensional model. Further data analysis of the scale revealed that the GPS is an adequate assessment tool for adults, with good psychometric characteristics and high internal consistency. Conclusion The model proposed in the current work leads to further refinements and enrichment of the construct of paranoia in different populations, allowing the assessment of three dimensions of paranoia and the risk of clinical paranoia in a single measure for the general population.