Analysis of allelic loss in archival tumor specimens using array technology is constrained by the quality and quantity of available tissue. A prototype Affymetrix HuSNP array was shown to provide reliable and reproducible assessment of allelic imbalance for 440 single-nucleotide polymorphisms in frozen esophageal tumors (Mei-2000). However, the commercially available Affymetrix HuSNP array (1,494 single-nucleotide polymorphisms) has not been validated for the assessment of allelic imbalance in tumors processed by standard pathology methods. We tested the HuSNP assay in duplicate on breast specimens using both formalin-fixed and frozen, tumor and normal tissue taken from a single patient (16 arrays). We purified tumor cells using bivariate cytokeratin/DNA flow sorting; normal breast served as the constitutive normal. STR typing on three chromosomes validated regions of allelic imbalance. Allele calls from the HuSNP array averaged 95% reproducibility between duplicates and 94% concordance between the fixed and frozen samples. We also tested DNA from the same samples that had been subjected to whole-genome amplification (primer extension preamplification) before array analysis. Although overall signal intensities were lower, the data from this material was reproducible in duplicates and concordant between sample types at rates similar to those for genomic DNA. Results from genomic normal tissue DNA averaged informative (AB) calls at 379 loci over all chromosomes. Although data points were clustered and large segments of chromosomes were not informative by this technique, our data indicated that the Affymetrix HuSNP assay could potentially provide a low-resolution, genome-wide analysis of allelic imbalance in routinely processed pathology specimens.