OBJECTIVES: The study seeks to improve understanding of maternity health seeking behaviors in resource-deprived urban settings. The objective of this paper is to identify the factors which influence the choice of place of delivery among the urban poor, with a distinction between sub-standard and "appropriate" health facilities. METHODS: The data are from a maternal health project carried out in two slums of Nairobi, Kenya. A total of 1,927 women were interviewed, and 25 health facilities where they delivered, were assessed. Facilities were classified as either "inappropriate" or "appropriate". Place of delivery is the dependent variable. Ordered logit models were used to quantify the effects of covariates on the choice of place of delivery, defined as a three-category ordinal variable. RESULTS: Although 70% of women reported that they delivered in a health facility, only 48% delivered in a facility with skilled attendant. Besides education and wealth, the main predictors of place of delivery included being advised during antenatal care to deliver at a health facility, pregnancy "wantedness", and parity. The influence of health promotion (i.e., being advised during antenatal care visits) was significantly higher among the poorest women. CONCLUSION: Interventions to improve the health of urban poor women should include improvements in the provision of, and access to, quality obstetric health services. Women should be encouraged to attend antenatal care where they can be given advice on delivery care and other pregnancy-related issues. Target groups should include poorest, less educated and higher parity women.