The article describes trends in social class segregation between British secondary schools from 1984 to 1999, and includes "home international" comparisons of trends in England, Wales and Scotland. The analysis is based on comparable youth cohort datasets: the England and Wales Youth Cohort Study and the Scottish School Leavers' Survey. The measures of social class are derived from parents' occupational status and education, and thus provide different dimensions of segregation from those used in some earlier studies which focused on average free school meal entitlement. Segregation indices derived using standard methods include a measure of (un)evenness (S) and a measure of isolation (I). In addition, the variance ratio is derived using a multilevel statistical model, with the advantage that confidence intervals can be estimated. Differences in segregation between the three countries are small, but the indices provide evidence that segregation is consistently lower in Scotland than in England, and this is compatible with the view that the more comprehensive system in Scotland is associated with lower segregation. However, the picture for Wales is rather varied, and hampered by small sample size and changing sampling methods. In the 1990s, the trajectory of segregation in Scotland was different from that in England or Wales: it may have fallen slightly in Scotland but risen elsewhere. But gaps in the available data prevent us being sure about that. The segregation indices do not reveal clear upward or downward trends in the period since parental choice legislation was introduced. The following are appended: (1) Sample sizes per school; and (2) Multilevel model for variance ratio.(Contains 14 figures and 4 notes.)