Four techniques of administering peripheral parenteral nutrition (PPN) were examined prospectively to investigate the role of mechanical trauma in the development of infusion phlebitis. Patients in group 1 (n = 15) were fed via a standard 18-G Teflon cannula which was removed on completion of the infusion and was rotated to the contralateral arm every day. Group 2 patients (n = 15) had a similar catheter sited in each forearm simultaneously, with rotation of the side of infusion each day. Patients in group 3 (n = 17) had a 15-cm Silastic rubber catheter inserted into a forearm vein and a standard cannula sited in the contralateral forearm, with alternation of infusion each day. Those in group 4 (n = 13) had a fine-bore 23-G silicone catheter sited in one arm only. Patients in groups 1, 2 and 3 were fed over 12-h cycles and those in group 4 for a 24-h continuous cycle. A total of 408 patient-days of PPN were given. Mean duration of PPN in groups 1-4 was 7.5, 9, 5.5 and 5 days respectively. Infusion phlebitis was not recorded in patients who had a daily change of cannula (group 1), but occurred in four patients in group 2, eight in group 3 and eight in group 4. Phlebitis scores were 0, 9, 15 and 12 for groups 1-4 respectively. Severe phlebitis and line occlusion occurred more frequently in patients with a 15-cm catheter (group 3) and in those fed continuously over 24 h (group 4). These results suggest that mechanical trauma is an important factor in the aetiology of infusion phlebitis. This can be minimized by reducing the time for which the vein wall is exposed to nutrient infusion and by reducing the amount of prosthetic material within the vein.