OBJECTIVE: We sought to determine how often non-English-speaking (NES) persons are excluded from medical research. DESIGN. Self-administered survey. PARTICIPANTS: A Medline search identified all original investigations on provider-patient relations published in major U.S. journals from 1989 through 1991, whose methodologies involved direct interaction between researcher and subject (N = 216). Each study's corresponding author was surveyed; 81% responded. MEASUREMENTS AND MAIN RESULTS: Of the 172 respondents, 22% included NES persons; among these includers, 16% had not considered the issue during the study design process, and 32% thought including the NES had affected their study results. Among the 40% who excluded the NES (excluders), the most common reason was not having thought of the issue (51%), followed by translation issues and recruitment of bilingual staff. The remaining 35% (others) indicated that there were no NES persons in their study areas. CONCLUSIONS: NES persons are commonly excluded from provider-patient communication studies appearing in influential journals, potentially limiting the generalizability of study findings. Because they are often excluded through overnight, heightened awareness among researchers and granting institutions, along with the development of valid instruments in varied languages, may increase representation of non-English-speaking subjects in research.