In this paper an incremental method and an interactive tool to improve the performance of word alignment are presented. The most important factor in this proposal is to put a human in the alignment loop. This is achieved by using interactive word alignment, i.e. a word alignment system and a human that are collaborating in order to make word alignment as efficient and accurate as possible. The aim is full coverage alignments with high accuracy, as the quality of word alignments is crucial in many applications within corpus linguistics, contrastive linguistics and Translation Studies.