A collaborative research effort by several institutions investigated the dispersion of odors from a swine production facility. Trained human receptors measured downwind odor concentrations from four tunnel-ventilated swine finishing barns near Story City, Iowa, during twenty measurement events conducted between June and November 2004. Odor concentrations were modeled for short time steps using CALPUFF and AERMOD atmospheric dispersion models to compare predicted and measured odor levels. Source emission measurements and extensive micrometeorological data were collected along with ambient odor measurements using the Nasal Ranger® device (St. Croix Sensory, St. Paul MN), Mask Scentometer, odor intensity ratings, and air sample analysis by dynamic triangular forced-choice olfactometry (DTFCO). AERMOD predictions fit the odor measurements slightly better than CALPUFF with predicted concentrations being about half those predicted by CALPUFF. The Mask Scentometer and Nasal Ranger® measurements related best to the dispersion model output, and scaling factors of 3.0 for CALPUFF and 2.4 for AERMOD suggested for the Nasal Ranger® and 0.5 for the Mask Scentometer (both models). Measurements obtained using the Nasal Ranger®, Mask Scentometer, and odor intensity ratings correlated well to each other, had the strongest linear relationships, and provided slopes (measured: modeled) closest to 1.0. Converting intensity ratings to a dilution to threshold concentration did not correlate and relate as well, and this method was deemed less desirable for ambient odor assessment. Collection of ambient air samples for analysis in a olfactometry laboratory displayed poor correlations with other methods and should not be used to assess ambient odors.