Radio frequency identification (RFID)-enabled smart shelves have recently attracted enormous attention from both industry and academia. Retailers have an explosion of interest in deploying smart shelves for the automatic ordering and the reduction of inventory inaccuracy. This study explores the retailers' optimal policy for investing in a smart shelf inventory control system. Since the decision maker's risk attitude plays an important role in its investment in technology, we therefore specifically consider the retailers' risk attitude. The risk attitude is measured by the mean-variance analysis. We derive analytical expressions for the benefit of the smart shelf and specify the conditions under which retailers should invest in it. We also show the explicit relationship between the retailer's risk attitude and his optimal policy. Finally, we conduct a numerical analysis to provide managerial insights for retailers to invest in smart shelves.