High-throughput, human cell-based applications of RNA-mediated interference (RNAi) have emerged in recent years as perhaps the most powerful of a 'second wave' of functional genomics technologies. The available reagents and methodologies for RNAi screening studies now enable a wide range of different scopes and scales of investigation, from single-parameter assays applied to focused subsets of genes, to comprehensive genome-wide surveys based on rich, multiparameter readouts. As such, RNAi-based screens are offering important new avenues for the discovery and validation of novel therapeutic targets for several disease areas, including oncology. By enabling a 'clean' determination of gene function, that is the creation of direct causal links between gene and phenotype in human cells, RNAi investigations promise levels of pathophysiological relevance, efficiency, and range of applicability never before possible on this scale. The field of oncology, with its many assays using readily transfectable cell lines, has offered particularly fertile ground for showcasing the potential of RNAi-based genomics. However, like any other technology before it, RNAi is not without its own challenges, limitations, and caveats. Many of these issues stem directly from the choice of silencing reagent to be used in such studies, and the design of the overall screening strategy. Here, we discuss the basic design issues, potential advantages, and technical challenges of large-scale RNAi screens based on the use of chemically synthesized siRNA libraries.