There are many reasons to celebrate the recent discovery of RNA interference (RNAi). The finding that small double-stranded RNA molecules can silence gene expression has had a huge impact on both biological research and technological applications.

For example, it adds a further layer of complexity to our understanding of gene regulation; we have been alerted to the hundreds of small RNAs with potential regulatory roles in development and physiology; and new pathways and enzymes involved in RNA processing are rapidly emerging. RNAi has been exploited as an experimental tool to inactivate gene function, and has enabled large-scale genetic experiments to be performed more effectively in mammalian cells, when they had hitherto been feasible only in lower organisms. There is therapeutic potential for RNAi too, with the first clinical trials currently being put forward for regulatory approval. The Nature Insight in this issue reviews this exciting field.

But it also carries an important note of caution about RNAi as a technology, which is discussed in a review by Gregory Hannon and John Rossi (see page 371). RNAi has many advantages over other methods of gene inactivation in its apparent ease and specificity. But just as with these other methods, the specificity of silencing is not absolute and there is a danger of ‘off-target’ effects. Hannon and Rossi explain the potential pitfalls and offer some “rules of the road” for effective RNAi experiments.

We fully endorse their recommendations. Good experimental design and controls can alleviate many of the specificity problems, and we will work with peer reviewers to keep abreast of such requirements in this fast-moving field (see also Nature Cell Biol. 5, 489–490; 2003). In the future, a deeper understanding of the mechanisms of RNAi should allow better design of RNAi triggers and enable non-specific effects to at least be predicted, if not eliminated.