With the increasing popularity of GPS-equipped mobile devices and the development of geo-positioning technology, location-based services are attracting much attention, and thus spatial-keyword (SK) queries are becoming more and more important. Effective index structures for geo-textual data are the key for efficient processing of spatial-keyword queries. The use of in-memory indices reduces disk accesses and significantly improves query performance. In this paper, we design and implement in-memory spatial-keyword indices (IR-tree and its main variants) for the most common spatial-keyword queries (standard and group) to analyze and evaluate their performance. Finally, we conduct extensive experiments with real-world datasets to demonstrate the efficiency of the proposed in-memory index structures and the algorithms for the studied spatial-keyword queries. The main conclusion is that in-memory IR-tree obtains the best performing spatial-keyword index for all studied SK queries, due to its simplicity and lightweight compared to its variants.