RNAi has become a very versatile tool to silence gene expression in a variety of organisms, in particular when classical genetic methods are missing. However, the application of this method in functional studies has raised new challenges in order to design RNAi reagents that minimize false positives and false negatives. Because the performance of reagents cannot be validated on a genome-wide scale, improved computational methods are required that consider experimentally derived quality measures. In this chapter, we describe computational methods for the design of RNAi reagents for invertebrate model organisms and human disease vectors, such as Anopheles. We describe procedures for designing short and long double-stranded RNAs for single genes, and evaluate their predicted specificity and efficiency. Using a bioinformatics pipeline we also describe how to design a genome-wide RNAi library for Anopheles gambiae.