Most research results on the throughput of unit-load automated storage retrieval (AS/R) systems assume the randomized storage; that is, all slots In the entire rack are equally likely to be selected for the storage location. However, for the most existing AS/R systems which use Closest Open location storage policy, Park[6] mathematically showed that this assumption results in a significant difference from the actual performance of AS/R systems. In this study, we present more details about the performance of the unit-load AS/R systems by considering the rack utilization and rack density; and examine their impact on the throughput of AS/R systems by using computer simulation. A new storage policy named Shortest Cycle Location(SCL) is proposed. Simulation results indicate that the performance of SCL policy consistently outperforms the Closest Open location storage policy. We also define the "Extra-Cost Zone" compared to the "No-Cost Zone"[3] and explore its geometric and practical meaning.