There is now a substantial body of work that reveals persistent disadvantage faced by ethnic minorities in the UK across the last few decades in terms both of hyper-cyclical unemployment rates during economic downturns and ‘ethnic’ penalties. Recent research has indicated that unemployment has a greater scarring effect upon subsequent careers for UK ethnic minorities than for the majority population. Yet most extant research in this area is based on cross-sectional data. In this paper, we use the first six waves of Understanding Society to explore labour market dynamics, focusing on unemployment and earnings. Preliminary findings show that black Caribbean and black African men of working age face unemployment rates of around 25 per cent compared to the rate among white majority men of around 10 per cent; and working-age Pakistani and Bangladeshi women are three times as likely to be unemployed as their white peers (23% vs 7%). Mirroring patterns found in the Labour Force Survey, Pakistani and Bangladeshi men and women also earn substantially less than their white peers. Exploiting the longitudinal design of Understanding Society, this paper uses multilevel and Markov chain models to illuminate the processes of ethnic minority vulnerability to unemployment and lower earnings. It asks whether disadvantage is due to scarring effects from minorities’ inequalities in initial labour market position, or whether they are attributable to unmeasured characteristics of ethnic minorities. It also investigates whether there are patterns of adaptation over the career whereby minorities move from employment to self-employment. Finally, it explores heterogeneity in these processes, specifically whether they play out differently for minorities who start their careers in more advantageous positions, for those with higher-level qualifications, for migrants compared to UK-born, and for women compared to men.