This study aimed to measure abortion safety in Nigeria, Cote d'Ivoire, and Rajasthan, India using population-based abortion data from representative samples of reproductive age women. Interviewers asked women separately about their experience with "pregnancy removal" and "period regulation at a time when you were worried you were pregnant", and collected details on method(s) and source(s) of abortion. We operationalized safety along two dimensions: 1) whether the method(s) used were non-recommended and put the woman at potentially high risk of abortion related morbidity and mortality (i.e. methods other than surgery and medication abortion drugs); and 2) whether the source(s) used involved a non-clinical (or no) provider(s). We combined source and method information to categorize a woman's abortion into one of four safety categories. In Nigeria (n = 1,800), 29.1% of abortions involved a recommended method and clinical provider, 5.4% involved a recommended method and non-clinical provider, 2.1% involved a non-recommended method and clinical provider, and 63.4% involved a non-recommended method and non-clinical provider. The corresponding estimates were 32.7%, 3.0%, 1.9%, and 62.4% in Cote d'Ivoire (n = 645) and 39.7%, 25.5%, 3.4%, and 31.4% in Rajasthan (n = 454). Results demonstrate that abortion safety, as measured by abortion related process data, is generally low but varies significantly by legal context. The policy and programmatic strategies employed to improve abortion safety and quality of care are likely to differ for women in different abortion safety categories.