Scholars have found that poor English proficiency is negatively associated with wages using self-reported measures. However, these estimates may suffer from misclassification bias. Interviewer ratings are likely to more accurately proxy employer assessment of worker language ability. Using self-reported and interviewer ratings from the Multi-City Study of Urban Inequality, the authors estimate the impact of English proficiency on wages for men (n = 267) and women (n = 178) with Mexican ancestry residing in the Los Angeles metropolitan area. Use of interviewer proficiency ratings suggests a larger and more gradational language penalty as fluency falls, and women face a stronger penalty than their male counterparts. Moreover, controlling for worker accent and skin shade does little to alter these effects. (Contains 4 tables and 17 notes.)