The infiltration of vehicle emissions into a house from the attached garage was studied for 16 homes of differing designs using the same extensively characterized vehicle at each home. Before the in-home measurement program, the cold-start and hot-start tailpipe emissions and hot-soak evaporative emissions from a 1993 Buick Regal were measured using standard vehicle emissions measurement methods. The emissions were chemically characterized for methane, nonmethane hydrocarbons (NMHC), and carbonyl compounds. The in-home measurements occurred over two winter seasons (1997-1998 and 1998-1999) in Ottawa, Ontario, Canada. Samples of indoor air and garage atmosphere were characterized for carbon monoxide, carbon dioxide, methane, NMHC, and carbonyl compounds. During the second year, real-time measurements of carbon, carbon dioxide, and total hydrocarbons were made to determine when and for how long the emissions plume infiltrates the house. Chemical mass balance modeling results using 31 NMHC species suggest that between 9 and 71% of the concentrations measured in the house during the hot-soak test and between 13 and 85% of the concentrations measured in the house during the cold-start test could be attributed to vehicle emissions infiltrating from the garage. In contrast, increases in carbonyl compound concentrations caused by the vehicle were difficult to detect above the already significant levels found in the houses.