"Traditional toxicity testing methods rely on exposing whole organisms to chemicals and observing high-level responses such as mortality and reproduction. These methods are too slow, expensive, and ethically concerning to assess the tens of thousands of legacy and novel substances in need of testing, and do not provide much biological insight into the toxicity mechanism. There has been a growing push to develop new approach methods for toxicity testing that do not use animal exposures. One major objective of this movement is to conduct exposures in vitro, measure comprehensive molecular outcomes, and use the molecular data to predict and manage risk to whole organisms. This not only promises to make toxicity testing faster, less expensive, and more humane, but also promises to generate more informative data. Toxicogenomics, the measurement of 'omics data such as transcriptomics, proteomics, and metabolomics in the context of toxicology, is key to realizing this goal. However, toxicogenomics data is extremely complex and the lack of computing resources, programming skills, advanced statistical training, and knowledge of bioinformatics databases presents barriers to many researchers and regulators. The barriers are particularly pronounced for environmental toxicologists using ecologically relevant species, as there are few bioinformatics resources outside of a small number of model organisms. The objective of this thesis is to design new statistical methods and corresponding software for analyzing and visualizing toxicogenomics data to support decision-making in the context of environmental toxicology. Many of the additional barriers to using transcriptomics data in ecologically relevant non-model organisms are related to raw data processing and annotation. Chapter 3 presents a set of computational tools (EcoOmicsAnalyst, ExpressAnalyst, EcoOmicsDB) for producing and analyzing annotated counts tables from raw RNA-seq data from any species, regardless of whether there is a reference genome. Traditional transcriptomics results such as lists of impacted genes and pathways are difficult to integrate into regulatory decision-making processes. Chapter 4 presents EcoToxModules, custom gene sets for summarizing and communicating transcriptomics data that are focused on toxicologically relevant biological processes. Chapter 5 presents FastBMD, software for performing rapid transcriptomics dose-response modeling. Since dose-response results such as benchmark dose values and points-of-departure are already familiar to the toxicology community, this type of analysis is useful because it translates unfamiliar toxicogenomics data into the familiar dose-response framework. Finally, while the cost of acquiring whole-transcriptome data has decreased tremendously over the last few decades, it is still outside the scope of many research programs. EcoToxChips are qPCR arrays with 384 genes for six ecologically relevant species that address this issue because qPCR technology is cost-effective with widespread availability. Chapter 6 presents EcoToxXplorer, software focused on EcoToxChip data processing, analysis, and interpretation. Together, the chapters in this thesis aim to support the use of toxicogenomics data in decision-making processes by making it more usable and understandable to individual members of the toxicology community, while also enabling standardized workflows that can be easily accessed by many different people in different locations. One important aspect of this is that all software presented in this thesis are web-based, which means that they do not require users to have substantial computing resources or programming skills, and do not require local installation. Throughout statistical method and software development, a design-thinking framework was used to continuously obtain and incorporate feedback from a large group of stakeholders and potential end-users from academia, government, and industry"--