Despite important progress in understanding the molecular factors underlying the development of cancer and the improvement in response rates with new drugs, long-term survival is still disappointing for most common solid tumours. This might be because very little of the modest gain for patients is the result of the new compounds discovered and marketed recently. An assessment of the regulatory agencies' performance may suggest improvements. The present analysis summarizes and evaluates the type of studies and end points used by the EMEA to approve new anticancer drugs, and discusses the application of current regulations. This report is based on the information available on the EMEA web site. We identified current regulatory requirements for anticancer drugs promulgated by the agency and retrieved them in the relevant directory; information about empirical evidence supporting the approval of drugs for solid cancers through the centralised procedure were retrieved from the European Public Assessment Report (EPAR). We surveyed documents for drug applications and later extensions from January 1995, when EMEA was set up, to December 2004. We identified 14 anticancer drugs for 27 different indications (14 new applications and 13 extensions). Overall, 48 clinical studies were used as the basis for approval; randomised comparative (clinical) trial (RCT) and Response Rate were the study design and end points most frequently adopted (respectively, 25 out of 48 and 30 out of 48). In 13 cases, the EPAR explicitly reported differences between arms in terms of survival: the range was 0-3.7 months, and the mean and median differences were 1.5 and 1.2 months. The majority of studies (13 out of 27, 48%) involved the evaluation of complete and/or partial tumour responses, with regard to the end points supporting the 27 indications. Despite the recommendations of the current EMEA guidance documents, new anticancer agents are still often approved on the basis of small single arm trials that do not allow any assessment of an 'acceptable and extensively documented toxicity profile' and of end points such as response rate, time to progression or progression-free survival which at best can be considered indicators of anticancer activity and are not 'justified surrogate markers for clinical benefit'. Anticipating an earlier than ideal point along the drug approval path and the use of not fully validated surrogate end points in nonrandomised trials looks like a dangerous shortcut that might jeopardise consumers' health, leading to unsafe and ineffective drugs being marketed and prescribed. The present Note for Guidance for new anticancer agents needs revising. Drugs must be rapidly released for patients who need them but not be at the expense of adequate knowledge about the real benefit of the drugs.