The influences of kilovolt peak, milliamperage, reconstruction algorithm, targeting, and image magnification on thin-section (1.5-mm) computed tomography (CT) of the lung were studied in phantoms and patients. Retrospective targeted reconstruction (25-cm field of view) improved spatial resolution, while magnification did not. The bone reconstruction algorithm improved spatial resolution, compared with the standard algorithm, and in patients, bone algorithm images were considered superior to standard reconstructions. Although using the bone algorithm increases the visible image noise, increasing the kilovolt peak and the milliamperage can reduce this noise. However, in the patients studied, this reduction in noise was not usually judged as significant, except in the posterior, paravertebral part of the lung. An optimal technique for CT of the lung parenchyma should include thin-collimation, targeted scans reconstructed with a high-spatial-frequency algorithm and, in some patients, increased kilovolt peak or milliamperage.