The Knudson "two-hit" hypothesis has provided the rationale for studies that aim to identify tumor-suppressor genes by mapping regions of allelic loss (loss of heterozygosity, LOH). Although LOH has been found in practically all types of tumors, very few such projects have been successful in identifying their tumor-suppressor targets. The prime explanation for this failure is probably that researchers have, in general, been too credulous about the two-hit hypothesis, and too willing to ignore factors such as intratumor heterogeneity, contamination by normal cells, karyotypic complexity, homozygous deletions, gene dosage changes, and polymerase chain reaction artifacts. We suggest ways of minimizing these problems. Unfortunately, there is no guarantee that existing or newer methods, such as genomic microarrays and in situ single-nucleotide polymorphism analysis, will solve the difficulties of LOH analysis. The future prospects for LOH studies are, as ever, uncertain.