Optimal management of dyspnea in terminal cancer patients requires an understanding of the responsible pathophysiological mechanisms. This prospective study assessed visual analogue scales (VAS) of shortness of breath (SOB) and anxiety, bedside spirometry, maximum inspiratory pressure (MIP), chest radiography, arterial blood gases, hemoglobin, and electrocardiogram, if indicated, in 100 terminally ill cancer patients. Forty-nine percent of the patients had lung cancer. The median VAS scores for SOB and anxiety were 53 mm and 29 mm, respectively. Spirometry was abnormal in 93% of patients, with 5% having obstructive, 41% restrictive, and 47% mixed patterns. The median MIP was 16 cm H2O. Sixty-five percent of the patients had parenchymal or pleural involvement on chest radiograph. Twenty-nine percent had evidence of cardiac ischemia, recent or current myocardial infarction or atrial fibrillation. Patients had a median of five different abnormalities that could have contributed to their shortness of breath. Only anxiety (p = 0.001), a history of smoking (p = 0.02), and pCO2 levels were statistically significantly correlated with SOB VAS scores. The potentially correctable causes of dyspnea included hypoxia (40%), anemia (20%), and bronchospasm (52%). The finding of very low MIPs suggests severe respiratory muscle weakness may contribute significantly to dyspnea in this patient population. Further studies are needed to confirm this finding and characterize the underlying pathophysiology.