Assessment of perfusion and capillary permeability is important in both malignant and nonmalignant lung disease. Kinetic modeling of T(1)-weighted dynamic contrast-enhanced MRI (DCE-MRI) data may provide such an assessment. This study establishes the feasibility and interrelationship of kinetic modeling approaches designed to estimate microvascular properties in malignant and nonmalignant tissues of the lung. DCE-MRI data were acquired using a low molecular weight contrast agent with 4-sec temporal resolution in lung cancer patients. A model-free parameterization and three kinetic models of increasing complexity, each related to the classical Kety model, were applied. Comparison of an extended Kety model and the adiabatic approximation to the tissue homogeneity (AATH) model using Akaike's Information Criterion suggested that in most cases the best description of the lung tumor data is obtained using the AATH model. In the normal lung parenchyma the temporal resolution was insufficient to separate effects of flow and contrast agent leakage and in this case the extended Kety model yielded the best fit to the data.