We are developing a cell-based bioelectronic glucose sensor that exploits the multi-parametric sensing ability of pancreatic islet cells for the treatment of diabetes. These cells sense changes in the concentration of glucose and physiological hormones and immediately react by generating electrical signals. In our sensor, signals from multiple cells are recorded as field potentials by a micro-electrode array (MEA). Thus, cell response to various factors can be assessed rapidly and with high throughput. However, signal quality and consequently overall sensor performance rely critically on close cell-electrode proximity. Therefore, we present here a non-invasive method of further exploiting the electrical properties of these cells to guide them towards multiple micro-electrodes via electrophoresis. Parameters were optimized by measuring the cell's zeta potential and modeling the electric field distribution. Clonal and primary mouse or human β-cells migrated directly to target electrodes during the application of a 1 V potential between MEA electrodes for 3 minutes. The morphology, insulin secretion, and electrophysiological characteristics were not altered compared to controls. Thus, cell manipulation on standard MEAs was achieved without introducing any external components and while maintaining the performance of the biosensor. Since the analysis of the cells' electrical activity was performed in real time via on-chip recording and processing, this work demonstrates that our biosensor is operational from the first step of electrically guiding cells to the final step of automatic recognition. Our favorable results with pancreatic islets, which are highly sensitive and fragile cells, are encouraging for the extension of this technique to other cell types and microarray devices.