The human transcriptome is constituted of a great majority of intron-containing and a minority of intron-lacking mRNAs; given the different processing these transcripts undergo, they are expected to carry, intermingled with coding properties, very different editing information. Here we applied a computational approach to compare intronless and intron-containing coding sequences. Hexamer composition comparison allowed the definition of over- and under-represented motifs in intronless genes; surprisingly, experimental testing revealed that intron-lacking coding sequences are enriched rather than depleted in elements with splicing enhancement ability. Similarly, we show evidence that intronless transcripts display a significantly higher frequency of both shuttling and non-shuttling SR protein binding sites compared to intron-containing sequences. These observations suggest that SR proteins (and possibly other splicing factors) play a role in cellular processes distinct from splicing. (C) 2004 Elsevier Inc. All rights reserved.