The "off-target" silencing effect hinders the development of siRNA-based therapeutic and research applications. A common solution to this problem is an employment of BLAST, which may miss significant alignments, or an exhaustive Smith-Waterman algorithm, which is very time-consuming. A Comprehensive Redundancy Minimizer (CRM) approach was developed for mapping all unique sequences ("targets") 9-15 nt in size within large sets of sequences (e.g., transcriptomes). CRM outputs a list of potential siRNA candidates for every transcript of a particular species. These candidates could be further analyzed by traditional "set-ofrules" types of siRNA designing tools. For human, 91% of transcripts are covered by candidate siRNAs with kernel targets of 15 nt. The approach was tested with a collection of previously described experimentally assessed siRNAs, and a significant correlation was found between the efficacy and presence in the CRMapproved set (r = 0.215, p-value = 0.0001). An interactive database containing a precompiled set of all human siRNA candidates with minimized redundancy is available at http://129.174.194.243. Application of the CRMbased filtering minimizes potential off-target silencing effects and could improve routine siRNA applications.