Often a user is already familiar with a sequence family that he or she wants to align, so some knowledge about existing sequence homologies may be available. Such expert knowledge can be used to direct an otherwise automated alignment procedure. To facilitate the use of expert knowledge for sequence alignment, we proposed an anchored alignment approach where known homologies can be used to restrict the alignment search space. This can clearly improve the quality of the produced output alignments in situations where automatic procedures are not able to produce meaningful alignments. In addition, alignment anchors can be used to reduce the program running time. For the Hox gene clusters that we analyzed, the non-anchored version of DIALIGN produced serious misalignments. We used the known gene boundaries as anchor points to guarantee a correct alignment of these genes to each other.