A practice survey was performed in Tenon hospital on 396 consecutive patients treated for solid tumors during 4 weeks in november 2002. 33% of anticancer drugs were off label used. The wording heterogeneity of the different anticancer drugs approved labeling and the lack of anticancer drugs in a number of cancers can explain those results. On one hand, randomised comparative clinical trial, considered as the best level of evidence to obtain a label used, is not always possible in cancerology, especially for rare tumors. One the other hand, pharmaceutical firm are not obliged to asked a label used for an anticancer drugs in spite of high level of evidence. So, label used can not be the own references for anticancer drugs prescribing, therapeutic advanced can be realised and disseminated before their taking into account in the label used.