Background: Diaphragm position and shape on chest radiographs are routinely used as indicators of normal or abnormal lung volume. However, previous population studies of normal diaphragm position and shape frequently lack objective radiographic measurements and documentation of pulmonary function, and do not account for the observed variability. Study objectives: To determine the spectrum of diaphragm position and shape on chest radiographs using objective measurements, in patients with normal pulmonary function, and to identify the relationship of diaphragm position and shape to demographic variables and radiographic thoracic dimensions. Design: Prospective cross-sectional observational study. Setting: University hospital. Patients: One hundred fifty-three adults with normal FEV1 (≥ 80% predicted) and normal total lung capacity (80 to 120% predicted). Measurements and results: Diaphragm position and shape relative to anatomic landmarks were determined from posteroanterior and lateral chest radiographs. We used descriptive statistics to calculate normal values, and linear correlation, two-tailed t tests, and multivariate analysis to relate findings to age, weight, gender, and thoracic dimensions. The right hemidiaphragm dome was positioned at 9.7 ± 0.8 thoracic vertebral levels (body plus disk space) [mean ± SD] below the top of the first thoracic vertebra (range, 7.4 to 11.3 vertebral levels), and the left hemidiaphragm dome was positioned at 10.2 ± 0.8 vertebral levels (range, 8.1 to 11.8 vertebral levels). The right hemidiaphragm radius was 13.8 ± 3.8 cm (range, 7.2 to 29.6 cm), and the ratio between the height and the anteroposterior dimension of the right hemidiaphragm was 0.23 ± 0.05 cm (range, 0.08 to 0.36 cm). The diaphragm tended to be lower with higher age, lower weight, and smaller transverse and anteroposterior thoracic dimensions (r = 0.22 - 0.47, p < 0.05), and flatter (larger radius) with higher age, weight, transverse thoracic dimension, pack-years smoked (r = 0.32-0.42, p < 0.0001), and male gender (p < 0.0001). The tested variables accounted for approximately one third of the variability in diaphragm position and shape (multivariate R2 = 0.31 - 0.38). Conclusions: The substantial variability in normal diaphragm position and shape is related to weight, age, and thoracic dimensions. Consideration of these factors may be useful when evaluating chest radiographs.