Motivation: Application of phylogenetic footprinting techniques to the analysis of the bacterial regulatory signals results in generation of sets of conserved candidate sites found upstream of orthologous genes. The next step of such analysis should be clustering of sites corresponding to the same signal and thus likely being the binding sites for the same regulator. Results: Here we present an algorithm for clustering of candidate regulatory sites leading to identification of groups of co-regulated genes. Availability: Algorithm was realized using C++ language as console application for Win32. Source code is available upon request.