BACKGROUND: Chromosomal translocations t(11;22) (q24;q12) are characteristic of about 80-90 % of Ewing's sarcoma family of tumors [bone and soft tissue Ewing's sarcoma and peripheral neuroectodermal tumors (PNET)]. They generate ews/fli1 rearrangements showing great diversity in breakpoint exon combination. In about 5 % of Ewing's tumors, ews is fused to the erg gene at 21q22. The various chimeric proteins encoded may function as aberrant oncogenic transcription factors. These specific translocations can be used for exact molecular diagnosis in these poorly differentiated small round-cell tumors. Moreover, the prognostic relevance of different translocational variants has been previously suggested. Furthermore, the sensitive molecular detection of minimal metastatic and residual disease and its clinical significance can be evaluated. To address these questions more definitively in the large number of patients registered in multicenter studies, it is often necessary to access archival paraffin-embedded tumor tissue if no fresh or frozen tumor material is available for analysis by RT (reverse transcription)-PCR. Specific problems arise from formalin-fixed and paraffin-embedded tissue due to the degradation of RNA and insufficient extraction efficiency. Therefore, primer distance and product size are limited for successful PCR amplification. This conflicts with the requirement for identification of various possible exon combinations by PCR simultaneously using one single primer pair with larger distance. PATIENTS: We examined paraffin embedded soft part tumor tissue samples from 47 Ewing's tumor patients. Patients were treated according to either CWS (Cooperative Weichteilsarkomstudie, CWS-91 or CWS-96) or Euro-E.W.I.N.G. 99 therapy protocols. METHOD: We established a novel RT-PCR method, using 3 different exon specific sets of PCR primer pairs, selected according to the coding ews and fli1 nucleotide sequences (NCBI database), suitable for RT-PCR identification of variant ews/fli1 fusion transcripts in RNA isolated from formalin-fixed, paraffin-embedded tissue. For use in combination with ews -primer, an erg specific primer was selected to alternatively test for ews/erg fusion transcripts. As positive control for the integrity of isolated mRNA, we used the ubiquitously expressed gapdh transcript for RT-PCR amplification in each sample. RESULTS: In 31 cases (= 66 %) of 47 paraffin samples of Ewing's tumors analysed, gapdh control indicated adequate quality of RNA. In 16 cases no gapdh control fragment was amplifiable, nevertheless in 2 of these 16 samples distinct ews fusion products could be detected. In 23 cases we identified ews fusion transcripts. Thereof in 65 % ews exon 7 being fused to fli1 exon 6 (fusion type I), in 22 % to fli1 exon 5 (fusion type II). In 4 % each ews exon 10 being juxtaposed to fli1 either exon 6 or exon 5, respectively. An ews/erg fusion was detected in 4 % ( ews exon 7 fused to erg exon 6). In 10 samples, a gapdh fragment was amplified, but no ews/fli1 or - erg fusion transcript could be identified. The reference pathological review (I. L., Kiel, Germany) disproved the primary histopathology in 5 cases. CONCLUSIONS: Using our different sets of exon specific primer pairs, it was possible to detect 4 different breakpoints of ews/fli1 fusion transcripts and the ews/erg fusion by RT-PCR in RNA isolates from formalin-fixed, paraffin-embedded Ewing's tumor tissue. This method can be a very useful alternative in clinical situations (to ensure diagnosis and perform minimal metastatic and residual disease investigations) and in order to assess prognostic significance of translocation subtypes when no fresh tumor tissue is available.