The availability of several complete genome sequences and large numbers of expressed sequences, has led to the development of bioinformatic strategies for large-scale predictions of natural antisense transcripts. In the past two years, this has given at least 1,600 human pairs and 2,500 murine pairs of natural antisense transcripts. However, due to limitations in bioinformatic search tools, experimental validation of the predicted antisense transcripts is crucial. This has been performed only for a small fraction of the large number of predicted natural antisense transcripts. Additionally, bioinformatic approaches will not allow systematic identification of natural antisense transcripts in specific cell or tissue types. More recently, an experimental approach for identification of sense-antisense transcript pairs has been developed. The principle of the strategy is based upon the selection of double-stranded cDNAs as a result of hybridization between first-strand cDNAs that should be generated in the presence of actinomycin D. This experimental strategy allows systematic analysis of sense and antisense partners in any cell or tissue type.