Summary Background Quality of life instruments are increasingly important in assessing disease severity. However, some of these measurements have been developed on a more or less ad hoc basis. Although not well standardized, psychometric analyses can be used to retest, refine and shorten existing quality of life instruments. Objectives To psychometrically test and refine the Impact of Psoriasis Questionnaire (IPSO) and to compare the results of two different statistical approaches. Patients and methods Among 792 psoriasis patients who were included in the PUVA Follow-up Study, we used classical test theory (CTT) and Rasch analysis to test and optimize the IPSO. Thereafter, two shortened versions of the IPSO derived from these models were compared. Results CTT analyses of the original IPSO demonstrated suboptimal item performance for six of 16 items and inappropriate subscaling. In contrast to the original four subscales, factor analysis of the CTT version yielded three subscales (mental functioning, mental wellbeing and stigmatization). The Rasch approach, which included ordering of thresholds, differential item functioning and item fit, resulted in a unidimensional 11-item questionnaire. Although the two new versions of the IPSO shared only six items, both reflected the original IPSO well. However, several arguments such as lower correlation coefficients, higher Cronbach's α, ordered thresholds, unidimensionality and fewer differences among subgroups of patients suggested that the Rasch version of the IPSO may be the preferred instrument to use. Conclusions The IPSO can be improved and shortened and the Rasch-reduced version of this instrument is likely to assess the psychosocial impact of moderate to severe psoriasis on patients' lives best because it is a short, reliable and unidimensional measurement.