Abstract: Mechanistic relationships between putative biomarkers, clinical baseline and related predictors versus clinical outcome (efficacy/safety) are usually unknown, and must be deduced empirically from experimental data. Such relationships enable the implementation of a personalized medicine strategy in clinical trials to help stratify patients in terms of disease progression, clinical response, treatment differentiation, etc. The relationship between some biomarkers & clinical baseline predictors versus clinical outcome are typically stepwise or nonlinear, often requiring complex models to develop the prognostic and predictive signatures. For the purpose of easier interpretation and implementation in the clinic, defining a multivariate biomarker signature in terms of thresholds on the biomarker combinations would be preferable. In this talk, we present some methods for developing such signatures in the context of continuous, binary and time-to-event endpoints. Further, to evaluate the future sample performance of the biomarker signature, we proposed the concept of predictive significance via cross-validation. Results from simulations and case-study illustration will also be provided.