Understanding how the genome empowers the nervous system to express behaviors remains a critical challenge in behavioral genetics. The startle response is an attractive behavioral model for studies on the relationship between genes, brain, and behavior, as the ability to respond rapidly to harmful changes in the environment is a universal survival trait. Drosophila melanogaster provides a powerful system in which genetic studies on individuals with controlled genetic backgrounds and reared under controlled environmental conditions can be combined with neuroanatomical studies to analyze behaviors. In a screen of 720 lines of D. melanogaster, carrying single P[GT1] transposon insertions, we found 267 lines that showed significant changes in startle-induced locomotor behavior. Excision of the transposon reversed this effect in five lines out of six tested. We infer that most of the 267 lines show mutant effects on startle-induced locomotion that are caused by the transposon insertions. We selected a subset of 15 insertions in the same genetic background in autosomal genes with strong mutant effects and crossed them to generate all 105 possible nonreciprocal double heterozygotes. These hybrids revealed an extensive network of epistatic interactions on the behavioral trait. In addition, we observed changes in neuroanatomy that were caused by these 15 mutations, individually and in their double heterozygotes. We find that behavioral and neuroanatomical phenotypes are determined by a common set of genes that are organized as partially overlapping genetic networks.behavioral genetics ͉ epistasis ͉ sensorimotor integration ͉ startle behavior ͉ P-element insertional mutagenesis A major goal of behavioral genetics is to understand the relationship between the genome and the nervous system. From a neuroscience perspective, behaviors represent the ultimate expression of the nervous system. From a genetics perspective, behaviors are complex traits for which natural variation is caused by many interacting genetic variants, with allelic effects that depend on social and external environments, sex, and genetic background (1, 2). To date, most studies that have attempted to relate genetic variation to the neural regulation of behavior have adopted a ''one gene at a time'' approach. Such studies have made important contributions and generated significant insights. However, recent genomic approaches, in which candidate genes affecting behaviors are identified by comparing whole-genome expression profiles of genetically divergent strains, have implicated large numbers of coregulated genes affecting behaviors that have pleiotropic effects on other traits, and that would not have been a priori predicted to affect behavior (3-10). In the current study, we used quantitative genetic approaches to analyze the genes-brain-behavior relationship at the level of genetic networks rather than at the level of single genes.We used startle-induced locomotion to a mechanical disturbance in Drosophila melanogaster as a model behavior. Previously, we...