We provide an evaluation of query execution plans (QEP) in the case of queries with one or two spatial joins. The QEPs assume R -tree indexed relations and use a common set of spatial joins algorithms, among which one is a novel extension of a strategy based on an on-the- y index creation prior to the join with another indexed relation. A common platform is used on which a set of spatial access methods and join algorithms are available. The QEPs are implemented with a general iterator-based spatial query processor, allowing for pipelined QEP execution, thus minimizing memory space required for intermediate results.