The COVID-19 pandemic has had a range of negative social and economic effects that may contribute to a rise in mental health problems. In this observational population-based study, we examined longitudinal changes in the prevalence of mental health problems from before to during the COVID-19 crisis and identified subgroups that are psychologically vulnerable during the pandemic. Participants (N = 14 393; observations = 48 486) were adults drawn from wave 9 (2017-2019) of the nationally representative United Kingdom Household Longitudinal Study (UKHLS) and followed-up across three waves of assessment in April, May, and June 2020. Mental health problems were assessed using the 12-item General Health Questionnaire (GHQ-12). The population prevalence of mental health problems (GHQ-12 score ⩾3) increased by 13.5 percentage points from 24.3% in 2017-2019 to 37.8% in April 2020 and remained elevated in May (34.7%) and June (31.9%) 2020. All sociodemographic groups examined showed statistically significant increases in mental health problems in April 2020. The increase was largest among those aged 18-34 years (18.6 percentage points, 95% CI 14.3-22.9%), followed by females and high-income and education groups. Levels of mental health problems subsequently declined between April and June 2020 but remained significantly above pre-COVID-19 levels. Additional analyses showed that the rise in mental health problems observed throughout the COVID-19 pandemic was unlikely to be due to seasonality or year-to-year variation. This study suggests that a pronounced and prolonged deterioration in mental health occurred as the COVID-19 pandemic emerged in the UK between April and June 2020.