ImportanceBeyond traditional race and ethnicity demographic characteristics, additional discrete data variables are needed for informed health interventions in the US.ObjectiveTo examine whether COVID-19 vaccine uptake patterns and associated disease outcomes differ among language preference groups.Design, Setting, and ParticipantsA cohort study of 851 410 individuals aged 18 years or older in a large multispecialty health system in Minnesota and western Wisconsin was conducted between December 15, 2020, and March 31, 2022.ExposureSelf-identified language preference and limited English proficiency (LEP) as measured by interpreter need were used to create subgroups using US census categories and attention to capture languages known to represent refugee groups.Main Outcomes and MeasuresThe primary outcome was COVID-19 vaccination uptake rates and time to first vaccine. Secondary outcomes were rates of COVID-19–associated hospitalization and death.ResultsMost of the 851 410 participants (women, 493 910 [58.0%]; median age, 29 [IQR, 35-64] years) were US-born English speakers; 7.5% were born in other countries, 4.0% had a language preference other than English (LPOE), and 3.0% indicated LEP as measured by interpreter need. Marked temporal clusters were observed for COVID-19 vaccination uptake, hospitalizations, and deaths associated with primary series vaccine eligibility, booster availability, and COVID-19 variants. Delayed first-dose vaccine was observed with LPOE (hazard ratio [HR], 0.83; 95% CI, 0.82-0.84) and interpreter need (HR, 0.81; 95% CI, 0.80-0.82) compared with those with English language preference and proficiency. Patients with LPOE were approximately twice as likely to be hospitalized (rate ratio [RR], 1.85; 95% CI, 1.63-2.08) or die (RR, 2.13; 95% CI, 1.65-2.69). Patients with LEP experienced even higher rates of hospitalization (RR, 1.98; 95% CI, 1.73-2.25) and COVID-19–associated death (RR, 2.32; 95% CI, 1.79-2.95). Outcomes varied for individual language preference groups.Conclusions and RelevanceIn this study, delayed time to first-dose vaccine was associated with increased COVID-19 hospitalization and death rates for specific LPOE and LEP groups. The findings suggest that data collection of language preference and interpreter need provides actionable health intervention information. Standardized system-level data collection, including at a national level, may improve efficient identification of social groups with disproportionate health disparities and provide key information on improving health equity in the US.