Recent developments in genomics and proteomics enable the discovery of biomarkers that allow identification of subgroups of patients responding well to a treatment. Although trial designs using a biomarker to identify these subgroups have been proposed in the literature, methodological research into the design of biomarker trials is in its early stages. Currently available designs are restricted in the way that they typically evaluate only a single treatment or biomarker and the performance of only some of the designs has been studied in detail. The overall aim of this project is to develop and evaluate novel biomarker designs overcoming the above mentioned limitations. The main objectives are: First, to derive novel analyzing strategies maximizing the information obtained from the data. This will be achieved by efficiently using subgroup data and deriving optimal randomization rules. Second, to extend existing biomarker designs and to develop new designs allowing evaluation of several treatments and/or biomarkers while controlling the overall type I error rate. Multiple testing procedures like the Bonferroni or the gate keeping procedure will be evaluated. Third, to incorporate short-term endpoints allowing for early stopping for futility/efficacy. In many disease areas the primary endpoint is only observed after a long follow up. Hence, at the time of the interim analysis little data is available for the primary endpoint. Methods like the double regression which incorporates information on short-term endpoints in order to improve the estimate of the long-term endpoint offer one way of solving this problem. Fourth, to allow inclusion of patients whose biomarker cannot be assessed by, for example, adding another arm randomising these patients between the treatments. To maximise the impact of my research, I will work closely with statisticians, clinical oncologists, and clinicians. Results will be presented at conferences and published in scientific journals.