Our 2006 McKim Conference on Predictive Toxicology addressed a needed paradigm shift in regulatory agencies toward a hypothesis-driven approach to chemical testing requirements. Conference participants outlined three scientific barriers that impede more efficient use of testing. Predicting in vivo toxicity endpoints from QSAR and in vitro data is one of the barriers which can only be overcome with a unifying strategy to integrate chemical and biological effects models in a better safety assessment. The McKim Conference and research programs at the International QSAR Foundation for the next five years will be devoted to developing and demonstrating a unifying framework for predictive toxicology. Recent legislative initiatives are permitting data gaps in hazard assessment to be filled with QSAR estimates, provided the QSAR methods meet with regulatory acceptance. OECD forged a set of principles to aid in selecting reliable QSAR methods. However, many assessment experts remain skeptical over QSAR models and QSAR estimates. In part, this skepticism is due to a tendency among QSAR modelers to gloss over the complexity of biological processes involved in long-term in vivo toxicity endpoints. The result is models that attribute complex biological responses to chemical attributes that have little or no relationship to the behavior of the chemical. To improve acceptance, the International QSAR Foundation proposed a conceptual framework that separates QSAR models of chemical interactions with biological systems from biological models of the effects progression from the molecular level to the organism level. The 2008 McKim Conference will expand that framework with practical examples of how to improve the mechanistic transparency in predicting complex toxicological endpoints for drugs, pesticides and industrial chemicals. The 2008 McKim Conference agenda flows logically from the concept of molecular initiating events in which chemicals interact with biological targets and alter the normal function of the target. Predicting these molecular initiating events from chemical structure is the purpose of QSAR. Predicting the cascade of biological effects from the sub-cellular level to the cell, organ and whole animal responses requires better toxicity pathway models, not chemical interaction models. The McKim Conference is exploring ways of organizing molecular targets and associated toxicity pathways that link biological effects across traditional levels of biological organization. In 2008, we hope to illustrate the organization of molecular targets at the organelle level, and have selected mitochondria-mediated effects to illustrate the conceptual framework. We will also be presenting new software tools for delineating and visualizing toxicity pathways.