Abstract : The assembly of multiple sequence alignments is one of the most common tasks in biology. Interestingly, the methods used for reconstructing these alignments are merely heuristics, and more or less well characterized from a biological or a mathematical point of view. The recent increase in the volume of available data is stressing these methods to their limits and the time has come to explore new strategies for coping with the outcome of high-throughput projects (structure, genomics, mass-spec, etc…). During this talk, I will review the current challenges and limitations of existing methods. I will discuss in detail consistency based methods and show why they could constitute the framework of data integration, at all possible levels. Among other things, I will discuss the integration of alternative methods (mcoffee), the integration of sequence and structure based methods (Expresso) and the alignment of RNA sequences using secondary structure information. All the methods presented here are freeware open source that can be either directly accessed or downloaded from the T-Coffee web server (www.tcoffee.org).