The use of magnetic resonance (MR) images for estimating mean pulmonary artery pressure (PAP) was tested by comparing main pulmonary artery (MPA) and middescending thoracic aorta (AO) caliber in 12 patients with primary pulmonary hypertension (PPH) with measurements made in eight other patients who were observed for diseases other than heart disease (controls). The ratio MPA/AO and the ratios of vessel caliber normalized to body surface area (MPAI and AOI, respectively) were computed. The PAP was obtained in all PPH patients and compared with caliber measurements. The PPH MPA (3.6 +/- 0.8 cm) was significantly larger than the control MPA (2.9 +/- 0.3 cm, p = 0.02); the PPH MPAI (2.8 +/- 0.7 cm/M2) was significantly greater than the control MPA (1.7 +/- 0.2 cm/M2, p < 0.0001). Control AO (2.2 +/- 0.3 cm) was significantly greater than PPH AO (1.6 +/- 0.4 cm, p < 0.0001); there was no significant difference between control AOI (1.3 +/- 0.2 cm/M2) and PPH AOI (1.2 +/- 0.2 cm/M2, p = 0.25). The PPH MPA/AO (2.3 +/- 0.6) was significantly greater than the control MPA/AO (1.3 +/- 0.1, p < 0.0001); overlap between MPA in the two groups was eliminated by indexing values to AO caliber (MPA/AO). Among PPH patients there was strong correlation between PAP and MPA/AO (PAP = 24 x MPA/AO + 3.7, r = 0.7, p < 0.01). Increased MPA/AO denotes the presence of pulmonary hypertension and may be used to estimate PAP.