The development of indicators and methods of collecting quantitative data on violence against women is central to both robust evaluation of policy developments and to the development of explanations. Three issues are addressed here: first, dilemmas in developing indicators based on consistent definitions; second, developments and remaining challenges in survey methodology; third, developments and gaps in collecting data from routine administrative sources. Indicators of violence against women need to capture the extent, as measured by both the rate of prevalence and the number of incidents, to measure severity by including injury levels, and to distinguish between acts carried out by intimate partners, other family or household members, and others. In the development of survey methodology, priorities are a comprehensive sampling frame and a high response rate, so as to include the more marginal groups of women who are especially likely to be at risk. In the delivery of the questionnaire, the methodological priority is that of confidentiality from both other household members and the interviewer in order to facilitate disclosure. Some public services collect some information that pertains to domestic violence, but this is rarely in a form relevant to policy. Relatively small additions to the data that is already collected would make it considerably more useful. However, such data does not give a guide to the actual level of violence against women, since many survivors do not report to authorities or use public services. Only survey data can provide a reliable guide to the extent of gender-based violence.