BACKGROUND: There is a growing global awareness of the psychological consequences of long COVID, supported by emerging empirical evidence. However, the emergence and long-term trajectories of psychological symptoms following the infection are still unclear. AIMS: To examine when psychological symptoms first emerge following infection with SARS-CoV-2 and the long-term trajectories of psychological symptoms comparing long- and short-COVID groups. METHOD: We analysed longitudinal data from the UCL COVID-19 Social Study (March 2020 to November 2021). We included data from adults living in England who reported contracting SARS-CoV-2 by November 2021 (n = 3115). Of these, 15.9% reported having had long COVID (n = 495). They were matched to participants who had short COVID using propensity score matching on a variety of demographic, socioeconomic and health covariates (n = 962 individuals with 13 325 observations) and data were further analysed using growth curve modelling. RESULTS: Depressive and anxiety symptoms increased immediately following the onset of infection in both long- and short-COVID groups. But the long-COVID group had substantially greater initial increases in depressive symptoms and heightened levels over 22 months follow-up. Initial increases in anxiety were not significantly different between groups, but only the short-COVID group experienced an improvement in anxiety over follow-up, leading to widening differences between groups. CONCLUSIONS: The findings support work on the psychobiological pathways involved in the development of psychological symptoms relating to long COVID. The results highlight the need for monitoring of mental health and provision of adequate support to be interwoven with diagnosis and treatment of the physical consequences of long COVID.