In this paper, we use data from the Labour Force Survey (LFS) to document the changing nature and characteristics of the self-employed in the UK between 2001 and 2020. A particular focus is on how self-employment has changed under COVID-19 relative to paid employment. Overall, there has been a sharp drop in the number of people self-employed and results from the longitudinal LFS indicate a marked increase in individuals reporting a move from self-employment to employment in 2020. The fact that there was no accompanying increase in the number of people changing jobs at this time may lend support to the hypothesis (Leaker, 2021) that once self-employed workers who paid themselves through PAYE realised that they were eligible for employee furlough payments through the Coronavirus Job Retention Scheme they began to self-identify as employees despite no genuine change in their status. An econometric model is used to estimate the relative impact of the pandemic on self-employed people. The results suggest, first, that the self-employed were harder hit in respect of the probability of remaining in work. However, this relative effect was arguably quite small, at just 0.6 percentage points, although it was roughly twice that for certain occupations (managers (directors) and senior officials; associate, professional and technical) and for those working in banking, finance & insurance etc. For comparison, we show that it is a similar relative impact to that of the Great Recession. A more notable impact is found on the second outcome considered: hours worked. The pandemic reduced hours worked among self-employed people more than among employees. The relative impact of 3.7 hours per week reflects both the reduction in the proportion in work and reduced hours among those remaining in work. Again, there was considerable variation across individuals, with stronger effects among men, non-whites and those in their mid-forties, and a mixed pattern of impact variation by occupation and industry. The third outcome is the probability of working zero hours for those in work. The results suggest the pandemic increased this probability by 4.7 percentage points more among the self-employed than among employees.