Survey interview languages not only determine who is eligible to participate in certain surveys but may also influence survey estimates. We examined potential biases arising from the exclusion of linguistic minorities using the 2003 California Health Interview Survey (CHIS), a multilingual survey. Although 88.3 percent of the interviews were conducted in English, results show that surveys using only English are likely to produce large differences in estimates. When survey interviews are conducted in English and Spanish, the overall difference may disappear, but the survey results for Asians remain unchanged. This implies that surveys conducted only in English may not adequately represent the general population and differences in estimates for Asians may exist even if English and Spanish are used.