Inactivation of tumour suppressor gene(s) (TSGs) on 3p appears to be a critical event in the pathogenesis of clear cell renal cell carcinoma (CC-RCC). Analysis of loss of heterozygosity (LOH) in sporadic RCC samples has implicated roles for TSGs in three specific regions of 3p in RCC development: (1) 3p12-p14, which includes the breakpoint of the familial t(3;8) constitutional translocation involved in hereditary RCC development and a recently cloned putative TSG, the FHIT gene: (2) 3p21.2-p21.3, a common region of deletion in many cancers including lung; and (3) 3p25-p26, which contains the von Hippel-Lindau (VHL) disease TSG. We and others have shown that most primary sporadic CC-RCCs contain somatic VHL gene mutations, clearly implicating inactivation of the VHL gene in the pathogenesis of CC-RCC. It is not known if CC-RCC without VHL gene mutations have alternative mechanisms of VHL gene inactivation or result from an alternative non-VHL pathway to RCC, e.g., inactivation of TSGs in 3p12-p21. We and others have reported hypermethylation and silencing of the VHL TSG in RCC from patients with VHL disease and in CC-RCC cell lines. However, the incidence and specificity of VHL methylation in primary sporadic RCC has not been defined. Therefore, we analysed methylation of the VHL, CDKN2, MYC, and H19 genes in primary RCC samples. Hypermethylation of the VHL promoter region was detected in 11% (11/99) of the primary RCCs analysed. In 10 of these tumours, there was no evidence of concomitant VHL gene mutation. VHL methylation was specific to CC-RCC (15%, 7/45) but was not detected in any non-CC tumours (n = 16). None of the 11 RCCs methylated at VHL had evidence of methylation at either CDKN2 or MYC (methylation at CDKN2 was, however, detected in 3%, or 1/33, of RCCs without VHL methylation). A normal methylation pattern at H19 was demonstrated in the three RCCs with methylated VHL analysed. Previous studies have suggested that, in addition to VHL, other 3p TSGs at 3p12-p14 and 3p21 may be involved in CC-RCC tumourigenesis. However, the interpretation of these studies has been difficult because information on VHL gene status has not been available for these data sets. Therefore, we investigated a subset of 55 sporadic RCCs (of known VHL gene methylation and mutation status) for LOH at polymorphic markers close to candidate TSG loci in the 3p14.2 and 3p21.2-p21.3 regions. Among tumours with LOH at one or more 3p markers, the incidence of 3p25 allele loss was higher in tumours with VHL alterations (mutation or methylation) than in those without. For tumours without detectable VHL alterations, the frequency of 3p14-p21 LOH was significantly higher than the frequency of 3p25-p26 LOH (93%, 13/14 vs. 43%, 6/14; P = 0.013), whereas, in RCC samples with VHL methylation or mutation, the frequency of 3p14-p21 LOH did not differ from that of sp25-p26 (72%, 18/25 vs. 59%, 13/22; P = 0.376). None of the 11 RCCs with 3p25 allele loss that were informative at 3p21 and 3p14 showed LOH at 3p25 only. These findings suggest that (1) VHL methylation is a specific and important event in the pathogenesis of CC-RCC; (2) in CC-RCC with 3p LOH but without VHL inactivation, mutations in TSGs at 3p14-p21 appear to have a primary role in tumourigenesis; and (3) inactivation of other 3p TSGs in addition to VHL may also be required for malignant transformation in tumours with VHL gene inactivation.