Extraordinary strain from COVID-19 has negatively impacted health care worker (HCW) well-being. To determine whether HCW emotional exhaustion has increased during the pandemic, for which roles, and at what point. This survey study was conducted in 3 waves, with an electronic survey administered in September 2019, September 2020, and September 2021 through January 2022. Participants included hospital-based HCWs in clinical and nonclinical (eg, administrative support) roles at 76 community hospitals within 2 large health care systems in the US. Safety, Communication, Organizational Reliability, Physician, and Employee Burnout and Engagement (SCORE) survey domains of emotional exhaustion and emotional exhaustion climate. The percentage of respondents reporting emotional exhaustion (%EE) in themselves and a climate of emotional exhaustion (%EEclim) in their colleagues. Survey items were answered on a 5-point scale from 1 (strongly disagree) to 5 (strongly agree); neutral or higher scores were counted as "percent concerning" for exhaustion. Electronic surveys were returned by 37 187 (of 49 936) HCWs in 2019, 38 460 (of 45 268) in 2020, and 31 475 (of 41 224) in 2021 to 2022 for overall response rates of 74.5%, 85.0%, and 76.4%, respectively. The overall sample comprised 107 122 completed surveys. Nursing was the most frequently reported role (n = 43 918 [40.9%]). A total of 17 786 respondents (16.9%) reported less than 1 year at their facility, 59 226 (56.2%) reported 1 to 10 years, and 28 337 (26.9%) reported 11 years or more. From September 2019 to September 2021 through January 2022, overall %EE increased from 31.8% (95% CI, 30.0%-33.7%) to 40.4% (95% CI, 38.1%-42.8%), with a proportional increase in %EE of 26.9% (95% CI, 22.2%-31.8%). Physicians had a decrease in %EE from 31.8% (95% CI, 29.3%-34.5%) in 2019 to 28.3% (95% CI, 25.9%-31.0%) in 2020 but an increase during the second year of the pandemic to 37.8% (95% CI, 34.7%-41.3%). Nurses had an increase in %EE during the pandemic's first year, from 40.6% (95% CI, 38.4%-42.9%) in 2019 to 46.5% (95% CI, 44.0%-49.1%) in 2020 and increasing again during the second year of the pandemic to 49.2% (95% CI, 46.5%-51.9%). All other roles showed a similar pattern to nurses but at lower levels. Intraclass correlation coefficients revealed clustering of exhaustion within work settings across the 3 years, with coefficients of 0.15 to 0.17 for emotional exhaustion and 0.22 to 0.24 for emotional exhaustion climate, higher than the .10 coefficient typical of organizational climate (a medium effect for shared variance), suggestive of a social contagion effect of HCW exhaustion. This large-scale survey study of HCWs spanning 3 years offers substantial evidence that emotional exhaustion trajectories varied by role but have increased overall and among most HCW roles since the onset of the pandemic. These results suggest that current HCW well-being resources and programs may be inadequate and even more difficult to use owing to lower workforce capacity and motivation to initiate and complete well-being interventions.