ABSTRACT Background: About 90% of women in Kenya report at least one antenatal care (ANC) visit yet maternal mortality rate remains high at 414 per 100,000 live births. Only 40% of childbirths occur in health facilities. A previous study of Nairobi slums in 2000 indicated that only 10.3% of women initiated ANC visits in the first trimester. High incidence of maternal deaths in Kenya especially among the very poor has been attributed to inadequate emergency obstetrical care. Decreasing numbers of women are initiating ANC within the first trimester and this may be affecting the ability of the health system to identify and cater for women whose health conditions can be effectively managed through ANC. This study aimed to determine the proportion of women initiating ANC in the first and last trimesters and the background characteristics associated with these women in two slums of Nairobi, Kenya. It also sought to determine if timing of initial ANC visit was associated with number of visits and choice of place of delivery in a slum setting. Materials and methods: This research report is a secondary data analysis of the World Bank funded Maternal Health Project conducted between 1st April and 30th June, 2006 by the African Population and Health Research Center. Participants were women 12 to 54 years, enumerated in the Nairobi Urban Health Demographic Surveillance System living in two slums of Nairobi, who had a pregnancy outcome between January 2004 and December 2005. Women 15 to 49 years were included in this analysis. Analysis of the data was done using STATA 9.2. Findings: Only 7.3% of women initiated ANC in the first trimester, with 52% making four or more visits. In the third trimester 22% of women initiated ANC. Although 97% of women reported receiving their first ANC from a skilled health professional, only 48.4% delivered in well equipped health care facilities. The median number of months pregnant at first ANC was six and median number of visits was four. Women who were most likely to initiate early ANC had secondary school or higher level of education (p=0.055) and were in a union (p=0.008). The least likely to initiate care in the first trimester were of minority ethnicity (0.011) and high parity (p=0.019). As educational level and wealth status rise, the likelihood of late ANC initiation declines. Women living with unemployed partners were less likely to initiate care in the first trimester compared to those living with employed partners (OR 0.2, p=0.046). Only women with educated partners initiated care during the first trimester. Women who initiated ANC in the first trimester were more likely to have 4 visits and more likely to deliver in appropriate facilities than those who initiated care in the third trimester. Those who initiated care to obtain an ANC card were less likely to have 4 visits than those who initiated care to verify that pregnancy was normal (OR 0.5, p=0.000). Women who initiated care in first trimester were 1.5 and 5.0 times more likely to deliver in good health facilities than those who initiated care in third trimester (p=0.040) and those who had no ANC (p=0.000), respectively. Conclusion: Women in Korogocho and Viwandani may have better chances of delivering in appropriate facilities if they have low parity and secondary level education. The presence of a partner with a means of steady income may also make it easier for women to access delivery care in good facilities. Interventions to improve the level of educational attainment among women and provide affordable family planning are necessary to increase early ANC attendance and subsequently delivery in well equipped facilities.