High-density oligonucleotide arrays can be extremely useful for identifying and quantifying specific targets (i.e., ribosomal RNA of microorganisms) in mixtures. However, current array identification schemes are severely compromised by nonspecific hybridization, resulting in numerous false-positive and false-negative calls, they lack an adequate internal control for assessing the quality of identification, and are dependent on amplification of specific target sequences which introduce biases. We have developed a novel approach for the routine quantification and identification of metabolically active microorganisms in mixed samples. The advantage of our approach over conventional ones is that it avoids designing, optimizing, validating, and selecting oligonucleotide probes for arrays; also, nonspecific hybridization is no longer a problem. The basic principle of the approach is that a fluorescence pattern of a mixed sample is a superposition of the fluorescent patterns for each target. The superposition can be quantitatively deconvoluted in terms of concentrations of each microbe. We demonstrated the utility of our approach by extracting rRNA from three microorganisms, making test mixtures, labeling the rRNA, and hybridizing each test mixture to DNA oligonucleotide (20-mers, n = 346,608) arrays. Comparison of known concentrations of individual targets in mixtures to those estimated by the solution revealed highly consistent results. The goodness-of-fit of the solution revealed that about 90% of the variability in the data could be explained. A new analytical approach for microbial identification and quantification has been presented in this report. Our findings demonstrate that including signal intensity values from all duplexes on the array, which are essentially nonspecific to the target organisms, significantly improved predictions of known microbial targets. To our knowledge, this is the first study to report this phenomenon. In addition, we demonstrate that the method is a self-sufficient analytical procedure since it provides statistical confidence of the quantification. © 2007 Elsevier B.V. All rights reserved.