The ability to utilize the RNA interference (RNAi) machinery for silencing target-gene expression has created a lot of excitement in the research community. RNAi in mammalian cells is achieved through introduction or expression of 21-23 bp small interfering RNAs (siRNAs) in cells or animals. Currently, there are six ways of producing siRNAs. siRNAs can be produced by chemical synthesis, in vitro transcription, or RNase III/Dicer digestion of long dsRNAs. Alternatively, they can be expressed in vivo from plasmids, PCR cassettes, or viral vectors that include a CMV or polymerase III (pol III) transcription unit. So far, these approaches have been used to create siRNAs for use in loss-of-function studies. However, it is clear that siRNAs also hold great promise as therapeutic tools. First, their activity seems to be very sequence-specific. Moreover, siRNAs could be modified in order to increase their stability and potency in vivo. Here, we will review the issues and findings related to siRNA design and production. Moreover, we will summarize new findings on siRNA specificity, modification, and delivery, which are critical to their use as therapeutic agents.