This study focuses on admissions criteria used for "comprehensive" secondary schools in London, UK. It was found that schools whose admissions were controlled by the local authority were more likely to report giving priority to children with medical/social needs and special educational needs than were schools that controlled their own admissions; the latter were more likely to report the use of various potentially "creaming" criteria. There was also more "selectivity" among London comprehensive schools with autonomy over admissions, with higher proportions using potentially selective admissions criteria than in the rest of England. Moreover, it was found that schools with responsibility for their own admissions had lower proportions of pupils with special educational needs and obtained higher scores in public examination "league tables" than schools whose admissions were controlled by the local authority. The findings suggest that some schools, although nominally "comprehensive", appear to restrict access to certain groups of pupils. (Contains 6 tables and 10 notes.)