A geographic information system was used to integrate information on the uptake of cervical cytology screening for general medical practices in Norfolk with other data on the characteristics of the practices and the population they served. Regression analysis was employed to explore the extent to which variations in non-response were associated with explanatory factors. Non-response to the earlier system of opportunistic screening was found to be associated with the size of practice, the presence of a female doctor, the District Health Authority in which the practice was located, the age structure of the practice population, its degree of socio-economic deprivation and levels of rural remoteness. Compared with the earlier system, the new population-based call and recall system was found to be reaching a larger proportion of the population at risk. The former relationships were weaker under the new system, but non-response was still significantly associated with both social and organisational factors.