NCHRP Project 8-78 is developing methods for Estimating Bicycle and Pedestrian Demand for Planning and Project Development. The envisioned end product is a practitioner guidebook on a range of methods to estimate non-motorized travel demand in response to land use, infrastructure and demographic factors. While one portion of the research team (Mark Bradley & John Bowman) are testing a tour-based modeling approach using Seattle data, the authors are leading a separate team that is attempting an accessibility-oriented approach based almost exclusively in GIS. Using travel data from the Metropolitan Washington (DC) Council of Governments’ 2008 regional travel survey, efforts are being made to explain choice of mode in relation to accessibility scores for each mode (walk, bike, transit, auto) and trip purpose. Potential trip attractions are represented through Dun & Bradstreet employment data, which reveal the exact location of each regional employer by size and function (SIC code). Access to these opportunities is determined through travel networks, which are unique to each mode. Bike or pedestrian-only links are added to the base street network to reflect the full array of choices, and mode-specific impedances are calculated for each link taking account of features such as slope, separation from traffic and crossing delay. The linked networks are also in GIS format (NAVTEQ), and so it is possible to use Network Analyst to build paths over the networks to determine the number and type of activities that can be reached by the respective mode within incremental levels of impedance (i.e., minutes of composite travel time). Through the cumulative opportunities density function thus formed, it is possible to compare modes using accessibility scores, and also create the means to estimate probability choice models for different trip purposes. The utility expression for each mode will be determined from the accessibility measure, which can then be used in a planning capacity to examine the effect of changes in land use or network speed/connectivity on mode choice probability, and thence, trips by mode.