This chapter describes several ligand-based models which we have developed for various human toxicity endpoints. We describe models for hERG, P450s, PXR and drug induced liver injury, with a particular focus on Bayesian models. These serve as a snapshot of the overall efforts to build computational models for toxicity now available in the literature. Such in silico models may help avoid compounds with toxicity or flag undesirable substructures prior to in vitro testing. Such models could serve to highlight adverse drug issues, to facilitate decision making without necessarily killing compounds immediately. We provide an overview of the literature and highlight cases where pharmaceutical companies such as GSK, AstraZeneca, Lilly, Pfizer, BMS, Roche and many others have published on their computational toxicity efforts. We conclude by suggesting that computational toxicology may be used more heavily than it has been in the past.