Fifty-five patients who had been started on continuous subcutaneous insulin infusion (CSII) more than 1 year previously were surveyed to determine the difference between those who had continued and those who had discontinued CSII. These retrospective results were used to determine characteristcs that predict a suitable candidate for CSII which could be tested in a future prospective study and would guide our clinical choice of candidates for CSII. Twenty patients had discontinued CSII (Group D) and 35 patients had continued CSII (Group C). Group D contained younger patients, more likely to be women who had had a shorter duration of diabetes, more frequent antecedent episodes of diabetic ketoacidosis (DKA), and more frequent doctor visits. The major rationale for starting CSII in Group D was poor control with ketoacidosis whereas Group C started CSII for poor control with hypoglycemia. The reasons for stopping CSII in Group D were interference with lifestyle (40%), needle sites abcesses (40%), psychological rejection (10%), and poor insulin absorption (10%). Group D patients had a significantly higher incidence of psychological rejection of CSII, dermatological problems, and needle site abcesses. For patients in Group D, glycosylated hemoglobin levels during CSII were similar to those prior to CSII. Group C patients had improved glycosylated hemoglobin levels and were more satisfied with CSII despite a slightly higher incidence of pump runaway. Group C patients had a trend toward more frequent hypoglycemia while on CSII; however, in patients who had reported severe hypoglycemia before starting CSII, there was a significant decline in the frequency of severe hypoglycemia. Even though they rejected the pump, Group D patients were found to have increased the frequency of insulin injections and home monitoring and to have decreased the frequency of DKA and doctor visits following CSII.