Psychological assessment of Hispanic populations are thwarted by the absence of clinical research on comparability of Spanish translations and the stability of their findings. In this study we examined the potential usefulness of the Spanish Personality Assessment Inventory (PAI; Morey, 1991) for Hispanic clients residing in Mexican American communities. We administered the Spanish version on two occasions to 48 monolingual clients and the Spanish and English versions to 21 bilingual clients. Results indicated that the clinical scales had a moderate to good correspondence for Spanish-English (M r = .72) and good test-retest reliability for Spanish-Spanish (M r = .79). Much more variation was observed for the validity scales and the treatment/interpersonal scales. Also more variability was observed in the convergence of elevations across administrations. Because of these mixed results, we discuss the potential usefulness of the PAI clinical scales to screen for major psychopathology.