The present study examines how learners' language mindsets (beliefs about the malleability of language intelligence) and achievement goals predict learners' preferences for different types of corrective feedback (CF). Questionnaire data were collected from 537 learners of Spanish as a foreign language at two North-American universities. Factor analytic and multiple regression results showed two clear patterns. A growth language mindset (the belief that language learning ability is malleable) predicted preferences for all types of CF whereas a fixed language mindset predicted a preference for conversational recast and absence of CF. A development-approach goal (concerned with improving one's language competence) positively predicted learners' preference for the explicit types of CF whereas a development-avoidance goal (concerned with maintaining one's second language competence) positively predicted the more implicit types of CF. Demonstration-approach (concerned with displaying superior competence) did not predict any CF type and demonstration-avoidance (concerned with avoiding the display of incompetence) predicted a preference for absence of CF. Using an overarching cost-value model, we discuss how learners' analysis of the self-presentation and ego costs as well as the value of different CF types have led to their preference for one versus another. Future directions and pedagogical implications are discussed.