Loss of heterozygosity and allelic imbalance in tumors are usually detected by either radioactive labeling of PCR products with subsequent scoring of autoradiographs or by a semi-quantitative fluorescence-based protocol. Polymorphic microsatellite loci are the most common marker type used in these studies. Even though no consensus exists as to how to evaluate such data, results are often compared directly between studies applying the two different protocols. In the present study, we analyzed twice by each protocol three loci in 60 blood/tumor pairs, finding good correlation between the results obtained by the two methods. However, a higher sensitivity and the possibility to correct for stutter peaks were among several advantages inherent in the fluorescence labeling approach. In addition, we determined the cut-off level for allelic imbalance scoring by the fluorescent primer protocol, by repeated analysis of 485 constitutional heterozygous genotypes at 20 different dinucleotide repeat loci. Based on the standard deviation, we found that allelic imbalance should be scored whenever the peak height of one allele in tumor DNA is reduced to less than 0.84 of its value in constitutional DNA, relative to the other allele. Applying this cut-off value, more imbalances are detected than by the visual scoring of autoradiographs. Our data therefore suggest that a lower threshold value (0.75) must be used when results from both fluorescent and radioactive assays are compared.