Objective. - To determine how often interpreters were used for Spanish- speaking patients, patients' perceived need for an interpreter, and the impact of interpreter use on patients' subjective and objective knowledge of their diagnosis and treatment. Design. - Cross-sectional survey. Setting. - Public hospital emergency department. Patients. - A total of 467 native Spanish-speaking and 63 English-speaking Latino patients presenting with nonurgent medical problems. Main Outcome Measures. - Patients' report of whether an interpreter was used, whether one was needed, self-perceived understanding of diagnosis and treatment, and objective knowledge of discharge instructions. Results. - An interpreter was used for 26% of Spanish-speaking patients. For 52%, an interpreter was not used but was not thought to be necessary by the patient. A total of 22% said an interpreter was not used but should have been used. When both the patient's English and the examiner's Spanish were poor, an interpreter was not called 34% of the time, and 87% of the patients who did not have an interpreter thought one should have been used. Nurses and physicians interpreted most frequently (49%), and professional interpreters were used for only 12% of patients. Patients who said an interpreter was not necessary rated their understanding of their disease as good to excellent 67% of the time, compared with 57% of those who used an interpreter and 38% of those who thought an interpreter should have been used (P<.001). For understanding of treatment, the figures were 86%, 82%, and 58%, respectively (P<.001). However, when objective measures of understanding diagnosis and treatment were used, the differences between these groups were smaller and generally not statistically significant. There were no differences between English-speaking Latinos and native Spanish-speakers who said they did not need an interpreter. Conclusions. - Interpreters are often not used despite a perceived need by patients, and the interpreters who are used usually lack formal training in this skill. Language concordance and interpreter use greatly affected patients' perceived understanding of their disease, but a high proportion of patients in all groups had poor knowledge of their diagnosis and recommended treatment.