Many publications have documented loss of heterozygosity (LOH) on many different chromosomes in a wide variety of tumours, implicating the existence of multiple tumour suppressor genes (TSGs). Knudson's two-hit hypothesis predicts that these LOH events are the second step in the inactivation of both alleles of a TSG. However, to date the number of TSGs identified that are inactivated mainly at the somatic level in cancers and are not inherited has remained disappointingly small. Here we postulate that the accurate mapping of LOH events in a series of tumours to define a common LOH region is greatly confounded by deficient LOH detection, genetic instability and intertumour heterogeneity. Finding the TSGs in chromosomal regions of frequent LOH might require 'brute-force' genomic approaches.