The flow rate accuracy and flow continuity of the Homepump, Infusor, Intermate, Paragon, Sidekick, and Springfusor 10 flow-restricting infusion devices were studied. Three of each device were tested in a temperature-controlled cabinet at the manufacturer-specified operating temperature and at 20 and 30 degrees C. The flow rates used were 100 ml/hr and the rate that would provide a 24-hour delivery of fluid, except in the case of the Springfusor 10, which was tested at the maximum and minimum flow rates. Flow rate was measured gravimetrically at 30-second intervals. The endopoint of infusion was defined as the start of the terminal-phase decline. The effect of refrigerated storage was studied by using the Intermate as an example of the elastomeric devices tested. All devices exhibited a variable flow profile during infusion except for the Paragon, which had a near-constant flow throughout. The average error in flow rate was within the manufacturer's specifications when the devices were used under the manufacturer-specified operating conditions for all devices except for the Sidekick. The definition of the endopoint made little difference in the flow rate results for the Infusor and Springfusor 10 but significantly affected the flow rate results for the other devices, for which there was a substantial terminal phase with a prolonged period of reduced flow rate. The mean flow rate, error, and coefficient of variation for the Intermate devices stored under refrigeration were significantly different from those for the devices not refrigerated before use. When operated as recommended by the manufacturer, five of six devices had a flow rate error within the manufacturer's specifications; the results were affected by endpoint definition.