The present study examines how Spanish heritage speakers in the United States perceive and produce Spanish lexical stress. Although Spanish heritage speakers are in general native bilinguals of Spanish and English, since they are mostly English-dominant, the present study intends to see whether these speakers have difficulty in perceiving stress contrasts in Spanish like what has been found with English second language (L2) learners (Saalfeld, 2009). Moreover, given that Spanish heritage speakers speak Spanish far less frequently than they hear it (Hakimzadeh and Cohn, 2007), the present study also examines whether a discrepancy is found between their perception and production of Spanish lexical stress. Spanish disyllabic stress minimal pair words (e.g., páso 'I pass' vs. pasó 'he/she passed') were used to examine whether Spanish heritage speakers are able to perceive and produced these stress contrasts in a target-like manner. Their performance was compared to those of English L2 learners and Spanish monolinguals. Results showed that Spanish heritage speakers did not differ greatly from Spanish monolinguals when they perceived Spanish lexical stress, unlike English L2 learners who had significantly lower accuracy than Spanish monolinguals. However, with regard to production, Spanish heritage speakers performed in a non-target-like manner, particularly in cases when the stress was in the penultimate syllable (i.e., páso-type words). Similar patterns were found in the production of English L2 learners. The findings of the present study suggest that there may be a discrepancy between heritage speakers' perception and production of the heritage language, which may be due to their asymmetrical use of the heritage language. The present study also implies that heritage speakers' production of the heritage language may be more affected by their dominant language (i.e., English) than their perception.