INTRODUCTION: Literature controversies exist regarding the prognostic value of VHL mutations. The objective was to compare paraffin-embedded and frozen section specimens for VHL mutations detection and to evaluate the reliability of DNA analysis in formalin-fixed tissues. METHODS: Seventy-six patients with clear cell renal cell carcinoma (RCC) previously assessed for VHL status from frozen samples were included. Seventy-three tumor samples were known to be mutated for VHL. DNA was extracted and an electrophoresis was performed to determine DNA quality. The whole coding sequence was synthesized by double PCR amplification followed by sequencing. Sequencing results were compared with those previously determined from frozen samples. RESULTS: DNA could be extracted from the 76 paraffin samples. DNA quality was highly degraded and significantly less amplified by PCR in 34.2%, resulting in no sequence available for analysis in 57.7% and discordance with frozen samples in 42.3% of the cases respectively. VHL mutations were found in 52.1% of the whole paraffin samples whereas 98% were mutated; 72% could be sequenced, resulting in 69.1% of VHL mutations in this subset. Only half of observed mutations were fully consistent with frozen analysis in the 3 exons. Neomutations were found in 10.5% and 28.9% of known mutations in frozen samples were not detected in paraffin blocks. Only DNA quality significantly influenced PCR amplification and sequencing. CONCLUSION: Tumoral DNA extraction and VHL mutation analysis can be performed from formalin-fixed paraffin-embedded (FFPE) tissue in RCC. But mutations identified tissues are not strictly concordant with those from frozen analysis and therefore results obtained from FFPE samples should be interpreted with care.