This paper analyses the induced demand hypothesis using a disaggregate approach at the link level. A panel data-set of Minneapolis-Saint Paul highway network for the years 1980-98 is constructed. A model that predicts the traffic flow on the link in terms of vehicle kilometres travelled (VKT) based on the flow and capacity conditions existing on the link in the previous years is specified and estimated. The flow and capacity conditions existing on the identified neighbouring parallel links are also taken into account. Socio-demographic characteristics like population of the Minor Civil Division (MCD) to which the link belongs and the surrounding MCDs are also considered. The results indicate that capacity enhancements in the previous years, given by lane additions, have a positive and significant effect on the VKT of the link, confirming the induced demand hypothesis. The elasticities are lower than reported in previous research, indicating the importance of separating new construction from the expansion of existing links.