Because of how important walk access is for transit travel, service changes that affect walking distance, such as route or stop relocation, call for modeling at a fine enough level to accurately reflect the often arbitrary aspects of the access network and of demand distribution within a zone. Case studies of stop relocation in Boston and Albany demonstrate the feasibility of parcel-level modeling on the unabridged street network using an assessor’s database. Parcel-level demand is estimated by allocating observed on/off counts as a function of a parcel’s land-use type, size (e.g., gross floor area), and location factors. With actual land-use and street network data, we show how stop service areas can deviate substantially from the simple geometric shapes that follow from assuming airline or rectilinear travel, and demand distribution can be far from uniform within a zone. These factors can significantly favor particular transit stop locations.