Stock-outs are one of a retail chains’ biggest problems because they lead directly to lost sales, reduced profits, and the potential loss of customers. This research applied probit regression to determine the relationship between various stock-keeping unit (SKU) attributes and retail stock-out performance. The data sample came from a large grocery retailer in Serbia and included two high-risk product categories consisting of a total of 115 SKUs and 98 stores. For the identification of stock-outs, a perpetual inventory aggregation method was used. Regardless of the category observed, the variables that were identified as having a detrimental impact on stock-out performance include stock-out at the distribution center, promotion, and high sales speed. On the other hand, a beneficial effect in terms of a reduced number of stock-outs was observed when the ordering process was performed using an automated ordering system.