Abstract: Subject of this invention is a process for the chemical synthesis of polynucleotides having totally or partially random sequences, based on the utilization, as synthesis monomer units for the random sequence part, or presynthesized mononucleotides and dinucleotides. Said synthesis is carried out on separate supports, so that on each of those supports is alternated at least one reaction cycle wherein a mixture of said dinucleotides is bound, with at least one reaction cycle wherein a mononucleotide is bound, and that in a preferred embodiment at the end of the n cycles required for a codon synthesis, the supports are mixed and then redivided into one or more reaction containers. The resulting polynucleotides are such that, for the random sequence part, each trinucleotide unit is fit to match only a limited number of codons, predefined for each unity in number and sequence, and the genetic code degeneracy effects can thus be eliminated. FIG.