Objective. Although studies suggest that the earnings of limited-English-proficient (LEP) Hispanic men have recently improved relative to the English fluent, it remains unclear as to whether specific Hispanic groups experienced similar improvements. Methods. Using 1990 and 2000 U.S. Census data, this study employs regression, wage decomposition, and quantile regression analyses to examine how gender and Hispanic ethnicity relate to the LEP-earnings penalty. Results. The LEP-earnings penalty fell significantly for Mexican-American men between 1990 and 2000. However, additional results suggest that this penalty increased for Cuban-American men and women (and, to a lesser extent, for Mexican-American women). Conclusions. Expanding trade and ethnic networks as well as reduced statistical discrimination have not systematically benefited all LEP Hispanic populations. Therefore, policies designed to enhance English-language proficiency may yield heterogeneous socioeconomic outcomes along the ethnic, gender, and income class dimensions.