Recent successes and very serious problems (including removal from the market) of blockbuster drugs are cited as examples that point to the inability of clinical trials to pinpoint rare events in which a drug can have a rare but severe toxic effect on people and can also kill people. The pharmaceutical industry wants to avoid development of drugs for chronic ailments such as arthritis and obesity that may, within a few years of use, be found to be excessively dangerous. Predictive toxicology and toxicogenomics will lead research efforts to ensure that such problems of risk can be solved during the early stages of drug design. Vioxx, for instance, was approved by the Food and Drug Administration (FDA) in 1999 as a NSAID, but was pulled from the market voluntarily because a risk of heart disease was demonstrated with the use of Vioxx at high doses and for more than 18 months. Clinical trials for FDA approval usually are conducted for no more than 11 months. Answers needed include the reasons that one of several COX-2 inhibitors in the same class targeting the same enzyme can be more toxic than the others and whether or not it is possible to predict cardiovascular risk due to prolonged use of Cox-2 inhibitors or any other drug. The complexity of biological systems can provide some answers. A suggestion is made that lack of facts is not the problem, but that the inherent hierarchical organization of complex systems is the real roadblock to preventing accurate predictions of higher level functions (here, for how an organ responds to a drug) from molecular structures alone. Drug developers are advised to fully investigate and know competitors, targets, target distribution, pathways, drug interactions, and genes, among other facts. For the short term, risk can be lowered for a patient's response to a drug by knowing the patient's genetic background, and the better use of drugs can be obtained while improving the safety profile and bolstering public acceptance of the drug approval process. In the case of Cox-2 inhibitors, the problem is not acute toxicity, but instead long-term effects. A new challenge for predictive toxicology emerges because, unlike interventional medicine, the body responds to persisting conditions through adjustment of internal states. Examples of such adjustments are weight gain, insulin resistance, and drug tolerance.