The mature mRNA always carries nucleotide sequences that faithfully mirror the protein product according to the rules of the genetic code. However, in the chromosome, the nucleotide sequence that represents a certain protein is interrupted by additional sequences. Therefore, most eukaryotic genes are longer than their final mRNA products. The human genome project revealed that only a tiny portion of sequences serves as protein-coding region and almost one quarter of the genome is occupied by non-coding intervening sequences. The elimination of these non-coding regions from the precursor RNA in a process termed splicing must be extremely precise, because even a single nucleotide mistake may cause a fatal error. At present, two types of intervening sequences have been identified in protein-coding genes. One of them, the U2-dependent or major-class is prevalent and represents 99% of known sequences. The other one, the so-called U12-dependent or minor-class of introns, occurs in much lesser amounts in the genome. The basic problem of nuclear splicing concerns i/ the molecular mechanisms, which ensure that the coding regions are correctly recognized and spliced together; ii/ the principles and mechanisms that guarantee the high fidelity of the splicing system; iii/ the differences in the excision mechanisms of the two classes of introns. We are going to present models explaining how intervening sequences are accurately removed and the coding regions correctly juxtaposed. The two splicing mechanisms will also be compared.