The usage of conformational ensembles constitutes a widespread technique for the consideration of protein flexibility in computational biology. When experimental structures are applied for this purpose, alignment techniques are usually required in dealing with structural deviations and annotation inconsistencies. Moreover, many application scenarios focus on protein ligand binding sites. Here, we introduce our new alignment algorithm ASCONA that has been specially geared to the problem of aligning multiple conformations of sequentially similar binding sites. Intense efforts have been directed to an accurate detection of highly flexible backbone deviations, multiple binding site matches within a single structure, and a reliable, but at the same time highly efficient, search algorithm. In contrast, most available alignment methods rather target other issues, e.g., the global alignment of distantly related proteins that share structurally conserved regions. For conformational ensembles, this might not only result in an overhead of computation time but could also affect the achieved accuracy, especially for more complicated cases as highly flexible proteins. ASCONA was evaluated on a test set containing 1107 structures of 65 diverse proteins. In all cases, ASCONA was able to correctly align the binding site at an average alignment computation time of 4 ms per target. Furthermore, no false positive matches were observed when searching the same query sites in the structures of other proteins. ASCONA proved to cope with highly deviating backbone structures and to tolerate structural gaps and moderate mutation rates. ASCONA is available free of charge for academic use at http://www.zbh.uni-hamburg.de/ascona .