An experiment was carried out with 203 Japanese College students, designed to test the validity of Heider's, Newcomb's and Rodrigues' models of interpersonal balance. The results showed that Heider's and Rodrigues' Balance Dominant models are good predictors of how people rank the eight P-O-X triads in regard to their coherence; when pleasantness of such triads is the criterion for the rank-ordering of them, Newcomb's and (for females) Rodrigues' Agreement Dominant models are the best predictors, when no explicit assumption of future contact between P and O is made; if such an assumption is explicitly indicated, Rodrigues' Attraction Dominant model is the best predictor. The confirmation of the predictive value of the models in the conditions under which they were expected to hold, in a culture so different than those in which they have been tested during the last decades, lends strong support to the assumption that they are transculturally and transhistorically valid.