Background: Mortality rates peaked early in the COVID-19 pandemic and then declined. Possible explanations are pharmacological and non-pharmacological treatments, vaccines and changing demographics. We sought to evaluate temporal trends in clinical characteristics and survival of patients hospitalised with COVID-19 during the first two years of the pandemic in Denmark. Methods: In this observational study, we included all adults with COVID-19 consecutively admitted to three hospitals in Copenhagen, Denmark, from March 2020 through March 2022. The primary outcome was overall survival up to day 90 from admission. We used multivariable Cox proportional hazards models to estimate the association of survival within five consecutive time-periods, based on admission date, adjusted for baseline characteristics, vaccination status, remdesivir and dexamethasone treatment. Results: In 1630 included patients, the median age [IQR] was 68 [52, 79] years, 56.6% were men and 86.2% had comorbidity. Clinical characteristics changed over time. The crude 90-day mortality rate peaked in March–June 2020 with 28.9%, decreased from July 2020 to 17.5%, and increased again in January-March 2022 to 28.6%. Lower hazard ratios for death were observed in individuals admitted from July 2020 and persisted after adjusting for baseline characteristics. Adjusting for vaccination, remdesivir treatment and dexamethasone treatment attenuated the association in patients requiring low-flow oxygen. Conclusions: Our study suggests lower hazard rates for mortality in patients hospitalised with COVID-19 from July 2020 compared to March-June 2020, mainly driven by lower mortality in patients with a need of oxygen at baseline.