A study was undertaken to compare the use of three types of deep-breathing devices in patients undergoing upper-abdominal operations. Seventy-nine patients were divided into three groups, each receiving preoperative bedside testing of pulmonary function and instruction in the use of one of three randomly assigned deep-breathing devices thought to be representative of those currently available (Triflo II, Bartlett-Edwards Incentive Spirometer, or Spirocare). Repeat testing and instruction were provided daily during each of the first five postoperative days. There were few statistically significant differences in pulmonary function, vital signs and white blood cell count, and no difference in length of postoperative stay. No device was uniformly acceptable to patients, and none was used as frequently as recommended. When left at the bedside and only one daily reinforcement of instructions, the three devices showed no clinically important differences.