This study investigated psychometric properties of an instrument for assessing perceived occupational value, the 26-item OVal-pd. Data from 225 Swedish subjects with and without known mental illness were analysed regarding fit to the Rasch measurement model (partial credit model), differential item functioning (DIF), and functioning of the OVal-pd four-category response scale. The reliability (index of person separation, analogous to Cronbach's alpha) was good (0.92) but there were signs of overall and item level (six items) misfit. There was DIF between people with and without mental illness for three items. Iterative deletion of misfitting items resulted in a new 18-item DIF-free scale with good overall and individual item fit and maintained reliability (0.91). There were no disordered response category thresholds. These observations also held true in separate analyses among people with and without mental illness. Thus, the first steps of ensuring that occupational value can be measured in a valid and reliable way have been taken. Still, occupational value is a dynamic construct and the aspects that fit the construct may vary between contexts. This has implications for, e.g., cross-cultural research and calls for identification of a core set of culture-free items to allow for valid cross-cultural comparisons. Copyright © 2009 Informa UK Limited