Background: Infusion set function remains the limiting factor of insulin pump therapy due to nonmetabolic complications. Here, we tested an investigational extended-wear infusion set prototype with a soft, angled, wire-reinforced cannula with three additional side holes, and compared failure mechanisms and tissue response with a commercial Teflon control. Methods: A total of 48 Teflon and 48 prototype infusion sets were inserted subcutaneously every other day for 14 days in 12 swine and infused with dilute insulin. After two weeks, tissue around cannulas was excised, and occlusions, leaks, and kinks were determined. Tissue was processed and stained to assess the total area of inflammation (TAI) and the inflammatory layer thickness (ILT) around the cannulas. Data were analyzed using Fisher’s exact, analysis of variance-general linear model, Kruskal-Wallis, and post hoc tests. Results: On average, the TAI surrounding the investigational cannula was 52.6% smaller than around the commercial control. The ILT was 66.3% smaller around investigational cannulas. Kinks occurred in 2.1% (investigational) vs 32.4% (commercial) cannulas ( P < .001). There was no difference in occlusion alarms and leaks onto skin. Conclusions: The data suggest that the infusion set prototype elicits less inflammation over an extended wear time and is resistant to kinking, compared with a commercial Teflon device. This is consistent with previously published data on the impact of cannula material/angle on the inflammatory tissue response. We highlight the following important aspects of infusion set design: (1) secure skin adhesion, (2) reliable cannula insertion, (3) automatic removal of the stylet, (4) cannula material/design that resists kinking, and (5) minimization of local tissue inflammation.