Background Cross-sectional studies have shown that the COVID-19 pandemic has had a significant impact on the mental health of healthcare staff. However, it is less well understood how working over the long term in successive COVID-19 waves affects staff well-being. Aims To identify subpopulations within the health and social care staff workforce with differentiated trajectories of mental health symptoms during phases of the COVID-19 pandemic. Method The COVID-19 Staff Wellbeing Survey assessed health and social care staff well-being within an area of the UK at four time points, separated by 3-month intervals, spanning November 2020 to August 2021. Results Growth mixture models were performed on the depression, anxiety and post-traumatic stress disorder longitudinal data. Two class solutions provided the best fit for all models. The vast majority of the workforce were best represented by the low-symptom class trajectory, where by symptoms were consistently below the clinical cut-off for moderate-to-severe symptoms. A sizable minority (13–16%) were categorised as being in the high-symptom class, a group who had symptom levels in the moderate-to-severe range throughout the peaks and troughs of the pandemic. In the depression, anxiety and post-traumatic stress disorder models, the high-symptom class perceived communication from their organisation to be less effective than the low-symptom class. Conclusions This research identified a group of health service staff who reported persistently high mental health symptoms during the pandemic. This group of staff may well have particular needs in terms of the provision of well-being support services.