The authors explored four methods for estimating quantum noise in CT images of the lung, each based on a different definition of patient size (water-equivalent diameter) and the relationship between noise and diameter determined in water phantoms. The accuracies of the four methods were evaluated using an image-subtraction method as a gold standard. Noise estimates based on patient sizes derived from chest area, water-equivalent area, non-lung area, and water-equivalent path length had maximum errors of 229%, 93%, 34%, and 57%, respectively. Considering the magnitude of noise variation across the lung volume (approximately 30%), noise estimate based on non-lung area was reasonably accurate.