This ex post facto study analyzes the relationships between NEO-PI-R domains and facets and 14 MCMI-III personality disorders scales in a Spanish nonclinical sample (N = 674). It also aims to explore differences and similarities with the results obtained by Dyce and O'Connor in an American sample with the same instruments. As expected, facet-level factor analyses forced to five factors showed a pattern of relationships strongly similar between both samples, with a total congruence coefficient of 92, and acceptable factor congruence coefficients, except for the Openness factor (.68). In accordance with the predictions by Widiger and Widiger et al. percentages of significant correlations were around 60% in both samples, with most of them agreeing. The domain-level multiple regression analysis also revealed a great resemblance between both American and Spanish results, Neuroticism being the strongest predictor of personality disorders. More differences arose in the multiple regressions at facet-level, although the variance accounted for by included facets was practically the same as the domains. The cross-cultural validity of the predictive value of the NEO-PI-R over the MCMI-III personality disorders and the relative benefits of the facet-level analyses over domain-level analyses are discussed.