The research compares the employment prospects of disadvantaged social groups in Britain over the past 30 years. It uses data from the General Household Survey, conducted almost every year between 1974 and 2003, with a total sample of 368,000 adults aged 20 to 59. A logistic regression equation estimates the probability of having a job for each member of the sample, taking account of gender and family structure, disability, ethnicity and age (and controlling also for educational qualifications and regional unemployment rates). Net differences in employment probabilities are interpreted as employment penalties experienced by the social group in question. Some of these penalties have increased, and others have decreased, over the period.