Southern blotting is a widely used method of determining loss of chromosomal alleles in tumors, but cannot be used to analyze small biopsies and most fixed, embedded tissues. These problems preclude analysis of many surgical neuropathology specimens. We have employed a polymerase chain reaction assay for loss of heterozygosity (LOH) based on the (CA)n dinucleotide repeat polymorphisms found in abundance throughout the human genome. We compared this method to conventional Southern blotting in detecting LOH on chromosome 10 in gliomas. From tissue sections of 14 paraffin-embedded, formalin-fixed gliomas, we amplified the (CA)n repeat D10S89 locus and compared (CA)n repeat patterns between tumor DNA and constitutional DNA. Loss of one chromosome 10 (CA)n repeat allele was seen in informative glioblastomas that showed allelic loss by Southern blotting, but not in gliomas that had maintained both alleles. The (CA)n repeat method can be applied to small and fixed, embedded specimens, is rapid and simple to perform, and uses highly polymorphic probes. We suggest that (CA)n repeats are a less exclusive and more rapid means of studying LOH in brain tumors than Southern blotting, and will provide further mapping data for the identification of tumor suppressor genes integral to glial tumorigenesis.