This study assessed the psychometric attributes of the modified Parkinson Psychosis Rating Scale (mPPRS). In an attempt to improve scale's scaling assumptions and content validity, all types of hallucinations were rated and all items were scored based on intensity. The scale was cross-culturally adapted to four Latin American countries (Argentina, Brazil, Ecuador, and Paraguay). Acceptability, internal consistency, factor structure, convergent and known-groups validity, and precision (standard error of measurement, SEM) were explored. A total of 388 patients with PD were included in the study (age, 64.5 +/- 10.7 years; 59.8% males; PD duration, 8.2 +/- 4.9 years). The mPPRS was highly usable in terms of missing values generated and scores distribution (total computable scores, 99.7%, ceiling effect, <15%). Scaling assumptions were acceptable as noted by the range of item-total correlations (0.14-0.55, only one coefficient below 0.2). Internal consistency was adequate for research use (Cronbach alpha, 0.7). Factor analysis identified two factors that accounted for 58.5% of the variance. Low correlation coefficients were found with cognitive function (SCOPA-Cog) and disease severity (CISI-PD) (r(S) <or= 0.30), whereas correlation with psychosis were high (r(S) = 0.56). Known-groups validity analyses indicated a significant increase in mPPRS scores by Hoehn and Yahr stage (P < 0.001). The SEM value was 1.06. Overall, the results suggest that the mPPRS is a useful tool for evaluation of psychosis in PD. The results show that some psychometric properties of the mPPRS are satisfactory albeit there is room for the improvement of scale's content validity and internal consistency. (c) 2010 Movement Disorder Society.