Detailed infiltration and pressure measurements were made by Lawrence Berkeley Laboratories on four electrically heated homes in the Pacific Northwest. The primary purpose of the study was to investigate the impacts of wind, temperature and mechanical systems on infiltration in real homes, with a view toward resolving infiltration modeling problems raised in recent studies. The predictions of two natural infiltration models (LBL and AIM2) are compared in detail with one another and the measured data. The separate influences of stack- and wind-induced flow are analyzed. The LBL model predictions are 23--97% greater than the AIM2 model; the measured data fall within the range bounded by the two models. An improved method of calculating the height parameter for both models is proposed. A simple model is presented to incorporate the infiltration effects of exhaust and supply systems and unbalanced flows due to duct leakage. An unbalanced flow to the conditioned space induces approximately one-half of its magnitude in additional infiltration when it is small relative to natural infiltration. The measured data agreed closely with the theoretical model. The homes in this study were relatively tight, and the infiltration was dominated by the stack effect. The homes were typically well sheltered, andmore » the predicted infiltration due to wind was small. Forced-air distribution systems were investigated in detail. Air handlers and associated duct leakage can have large effects on living-zone infiltration rates. At two sites, the infiltration rate increased by more than 50% when the air handler was running. Duct leakage was found at all three sites with air handlers. Closing even a single bedroom door can cause a major increase in infiltration when the air handler runs. Pressures measured across a single closed door ranged from 2--6 Pa. At one site, closing the bedroom door more than doubled the added infiltration produced by the air handler. 19 refs., 21 figs., 31 tabs.« less