Acute lower respiratory tract illness is common among children seen in primary care. We reviewed the accuracy and precision of the clinical examination in detecting pneumonia in children. Although most cases are viral, it is important to identify bacterial pneumonia to provide appropriate therapy. Studies were identified by searching MEDLINE from 1982 to 1995, reviewing reference lists, reviewing a published compendium of studies of the clinical examination, and consulting experts. Observer agreement is good for most signs on the clinical examination. Each study was reviewed by 2 observers and graded for methodologic quality. There is better agreement about signs that can be observed (eg, use of accessory muscles, color, attentiveness; kappa, 0.48-0.66) than signs that require auscultation of the chest (eg, adventitious sounds; kappa, 0.3). Measurements of the respiratory rate are enhanced by counting for 60 seconds. The best individual finding for ruling out pneumonia is the absence of tachypnea. Chest indrawing, and other signs of increased work of breathing, increases the likelihood of pneumonia. If all clinical signs (respiratory rate, auscultation, and work of breathing) are negative, the chest x-ray findings are unlikely to be positive. Studies are needed to assess the value of clinical findings when they are used together.