Using software (PAT and LECTOR) developed for the creation of the electronic Oxford English Dictionary, strategies applicable to other large data bases were developed to analyse systemic sex-role stereotyping. These are applied to the OED database as a whole and to sub-files of gender-related definition or quotation text. The corpus is also studied manually. Software-based strategies including text searches for collocations with gender-specific pronouns and possessive adjectives produce interesting results which are tested against information in previous research. Stereotypes are found most frequently in quotation text, to a lesser degree in definition text.