Exposure to the butter-flavoring compound diacetyl was assessed in four microwave popcorn manufacturing plants in the United States during multiple surveys from 2005-2007. Personal, breathing zone samples were collected and analyzed using NIOSH Method 2557. Samples were collected at the lapel, outside the powered air-purifying respirator used by any worker entering the slurry room. The data were evaluated for similarity of exposure across job duties and resulted in two exposure groups: (1) Mixers, those who routinely work in the slurry room mixing vegetable oil, salt, and flavorings, and (2) Non-mixers (all other production jobs). From 639 samples collected during surveys, summary estimates of exposures were calculated as the arithmetic mean and median of the data and by the geometric mean of lognormally distributed results. For Mixers, the range of the arithmetic mean exposures across the four plants was 0.057-0.860, and geometric means ranged from 0.029 to 0.231. For Non-mixers, the range of arithmetic mean exposures was 0.014-0.074 ppm, and geometric means ranged from 0.001-0.018. Because of concerns regarding reliability of the method with increasing humidity, mean exposures were also calculated for samples collected when relative humidity was 30% or less. For this restricted set of measurements, geometric statistics always produced higher summary metrics, compared with the full data set; the direction of change using arithmetic statistics was not uniform, being higher at two plants and lower at two plants. Overall, 49.1% of the 639 values were below the limit of detection; in the lower humidity analysis, 25.6% of 246 results were below the limit of detection. At one plant, ventilation changes were made in the slurry room; based on 39 additional samples (not included in the 639 values reported here) collected using the same protocol after the engineering change, exposure decreased from 0.614 to 0.061 ppm. Further work at these facilities to better characterize diacetyl exposure was not possible, as substitution resulted in diacetyl being removed from the product.