We analysed householders' access to improved water for drinking and other domestic uses in five selected low-income urban areas of Accra, Ghana using a survey of 1,500 households. Our definitions of improved water were different from those suggested by the World Health Organization (WHO). The results revealed that only 4.4% of the respondents had access to improved drinking water compared to 40.7% using the WHO definition. However, 88.7% of respondents had access to improved water for domestic uses compared to 98.3% using the WHO definition. Using logistic regression analysis, we established that the significant determinant of householders' access to improved drinking water was income. However, for access to improved water for other domestic uses, the significant factors were education, income and location of the household. Compared to migrants, indigenous people and people from mixed areas were less likely to have access to improved water for other domestic purposes. For the analysis using the WHO definitions, most of the independent variables were not statistically significant in determining householders' access, and those variables that were significant generated parameter estimates inconsistent with evidence from the literature and anecdotal evidence from officials of public health and water supply companies in Ghana.