Based on a literature review and the SSG experience, the most important prognostic factors in high-grade osteosarcoma appear to be the presence of detectable metastases at diagnosis, tumour volume, old age, sex, histologic response, and possibly tumoral P-glycoprotein expression. However, for an adolescent patient with non-metastatic extremity disease, there is no consensus regarding prognostic factors at initial presentation, and currently there is thus no established method for dividing them into high- and low risk groups for the purpose of treatment differentiation. It should also be remembered that available prognostic factors have been identified only in a retrospective manner, following aggressive treatment of all patients. Thus patients in "favourable" prognostic groups may simply be patients who have had a good effect from aggressive treatment, and how they would have done with reduced treatment remains to be shown. Obviously the best method for prognostication would be the direct demonstration of micrometastatic disease in the lungs or in peripheral blood. In the relatively near future, this may become possible with immunoscintigrapy or immunohistochemistry utilizing monoclonal antibodies [29-31]. In Ewing's sarcoma, the most powerful factors indicating poor prognosis are metastases at diagnosis, poor histologic response, large tumour size and possibly pelvic localisation. There appears to be a somewhat better international consensus regarding prognostic factors in Ewing's sarcoma than in osteosarcoma. Although several studies have implemented intensified treatment for poor prognostic groups [8, 32], the role (if any) of high-dose treatment with stem cell rescue remains to be proven. The same factors are prognostic both for the development of metastases and local recurrence, but in addition, surgical treatment as opposed to radiotherapy appears to reduce local failure rate [12, 17, 33, 34]. As in osteosarcoma, the near future offers promise regarding the detection and quantification of micrometastatses and minimal residual disease, by means of PCR techniques recognizing specific genetic changes in the Ewing family of tumors [35].