Abstract Background The COVID-19 pandemic has further exposed inequities in our society, demonstrated by disproportionate COVID-19 infection rate and mortality in communities of color and low-income communities. One key area of inequity that has yet to be explored is disparities based on preferred language. Methods We conducted a retrospective cohort study of 164,368 adults tested for COVID-19 in a large healthcare system across Washington, Oregon, and California from March – July 2020. Using electronic health records, we constructed multi-level models that estimated the odds of testing positive for COVID-19 by preferred language, adjusting for age, race/ethnicity, and social factors. We further investigated interaction between preferred language and both race/ethnicity and state. Analysis was performed from October–December 2020. Results Those whose preferred language was not English had higher odds of having a COVID-19 positive test (OR 3.07, p