Artificial pancreas technology, involving "closed-loop" controls with real-time blood glucose monitoring, has been increasing in reliability as its potential for clinical use and application grows. One such device, based on this technology, is the STG-22 (Nikkiso Co., Ltd., Tokyo, Japan) artificial pancreas apparatus. In order to assess the reliability and accuracy of the device for measuring blood glucose, it is important to compare its readings to those obtained using a 'gold standard' method, such as the hexokinase method. Therefore, in the present study, canine blood [glucose] measurements using the STG-22 were compared to those obtained using a previously established commercial reagent, Quickauto-Neo GLU-HK. Furthermore, two different sample types (whole blood versus plasma constituent) were compared to determine which sample type results in more accurate and optimal readings with the STG-22. Given that the STG-22 was not primarily designed for canine blood samples, results for canine blood samples were not accurate. Measurements performed by the STG-22 with whole blood were significantly lower than reference [glucose] counterparts. Alternatively, an opposite trend was observed with plasma measurements that were significantly higher. A conversion format using the following formula, Hexokinase [glucose] = STG-22 [glucose] × 1.407 + 1.532, was observed with canine samples in our study.