Repeating fragments in biological sequences are often essential for structure and function. Over the years, many methods have been developed to recognize repeats or to multiply align protein sequences. However, the integration of these two methodologies has been largely unexplored to date. Here, we present a new method capable of globally aligning multiple input sequences under the constraints of a given repeat analysis. The method supports different stringency modes to adapt to various levels of detail and reliability of the repeat information available. © 2006 Wiley-Liss, Inc.