Successful ICT policy analyses and implementation rely on a sufficient knowledge of how and why consumers adopt new technologies. For instance, analysts of the digital divide may need to quantify the factors that are perceived to drive this ICT usage gap. This paper focuses on understanding the key factors involved in residential internet choice, and therefore the digital divide. Using survey data collected in the UK, internet choice structure is analysed using a discrete choice model that embodies both human characteristic and product attribute data. The benefit of this approach is that it is possible to statistically test if the factors of choice vary between dial-up and broadband services. A logit model is estimated and this model reveals that price and income elasticity effects vary systematically by household type. This finding will be important to market planners wishing to understand and quantify the impact of the assessed factors on digital division.