We have designed and evaluated the performance of a custom Affymetrix microRNA (miRNA) microarray that enables the analysis of 15,090 unique mature miRNAs that are predicted or biologically verified. The comprehensive probe content of the array was selected from multiple sources including the Sanger miRBase sequence database (version 8.0), published reports, and a proprietary database of predicted miRNAs. Most of the ~31,000 25-mer probes on the array were designed by selecting two overlapping oligonucleotide sequences that span the mature miRNA sequences and a portion of the flanking precursors (pre-miRs). We also included antisense control probes as well as probes that target the predicted hairpin loop region of the pre-miR for a subset of well-characterized human miRNAs. We evaluated the array performance using a model system of two total RNA reference standards isolated from different tissues that were mixed in calibrated ratios. The custom array performance was compared to an enhanced mirVana™ miRNA Bioarray, using the same calibrated mixtures of reference RNAs. We assessed the reproducibility, relative sensitivity, dynamic range and concordance between the results from the two platforms. The performance of the two platforms was similar, with both platforms reliably detecting differential miRNA expression. For those miRNAs that are represented on both platforms, there was a strong correlation of the log2 ratios of differential expression levels between the RNA reference standards. Many of the miRNAs with detectable expression signals were novel predicted miRNAs derived from bioinformatic exploration of human genomic sequence data. This provides initial evidence that some of the predicted miRNAs are expressed in tissues and, therefore, may function as regulators by mechanisms consistent with other well-characterized miRNAs. In summary, our data illustrate the technological and biological feasibility of the custom high-density miRNA microarray for the comprehensive identification and global expression profiling of novel miRNAs.