Many emerging applications of location-based services continuously monitor a set of moving objects and answer queries pertaining to their locations. Query processing in such services is critical to ensure high performance of the system. Observing that one predominant cost in query processing is the frequent accesses to the database, in this paper we describe how to reduce the number of moving object to database server round-trips by caching query information on the application server tier. We propose a novel-caching framework, named QACHE, which stores and organizes spatially-relevant queries for selected moving objects. QACHE leverages the spatial indices and other algorithms in the database server for organizing and refreshing relevant cache entries within a configurable area of interest, referred to as the cache-footprint, around a moving object. QACHE contains appropriate refresh policies and prefetching algorithms for efficient cache-based evaluation of queries on moving objects. In experiments comparing QACHE to other proposed mechanisms, QACHE achieves a significant reduction (from 63% to $99%) in database roundtrips thereby improving the throughput of an LBS system.