Background. Understanding the rising speed and magnitude of mental health problems during COVID-19 lockdowns is of particular value to epidemiologists, clinical practitioners, and policy makers. We examine the dynamic changes in depressive symptom severity and associated risk factors during the Shanghai COVID-19 lockdowns in spring 2022. Methods. A cross-sectional data was collected from an online survey conducted between April 15-17, 2022, after nearly one month of the Shanghai COVID-19 lockdowns. The survey included the full set of the Beck Depression Inventory-2, and asked the participants to retrospectively recall the severity of their depressive symptoms in three different time periods: early March (March 1-15), late March (March 16-31), and early April (April 1-15). The survey also included several sociodemographic and health-related questions in these time periods, including the duration of online shopping, work, physical exercise, sleep, and social communication during the lockdown. 652 citizens (212 males) in Shanghai participated in this study. Results. We found that the recalled mild depression (BDI-2 score > 13) increased from 10.1% in early March to 21% in late March, and further to 35.9% in early April. Regression analyses show that lockdown duration and physical restriction jointly contribute to worsening depression. In addition, sleep duration and social communication during lockdown were associated with the severity of depression symptoms. Conclusions. These results highlight the significant depression problems lockdown and call for special attention to early psychological intervention once lockdown is initiated. The limitations of retrospective and cross-sectional evaluations are also discussed.