Most cancer clinical trials with combination therapy in phase II have failed in phase III. One of the reasons for this failure is that no sufficient forethought was given to the interactions between the two (or more) agents.  Before embarking on a clinical trial with (say) two agents, one should address the following questions: If the two drugs are positively correlated at any dose amounts, how to achieve the same tumor volume reduction with minimal negative side-effects? If the two drugs are antagonistic at certain ranges of the doses, how to avoid these zones of antagonism? What schedule is most effective in reducing tumor burden? e.g. in which order to give the drugs? How to reduce drug resistance by the tumor cells? We use mathematical models to address these questions, and give several examples with different drugs.