The need to implement genetic resources management strategies within international networks is particularly acute for citrus. It appears to be an urgent priority to consider the significant loss of in situ diversity occurring in the Asian citrus native areas, due to environmental degradation, and biotic constraints. Technical and informative characteristics of STMS markers make them particularly adapted for both in situ and ex situ genetic resources management within networks. We are currently developing such markers in Citrus from a small insert genomic library of Cleopatra mandarin enriched for (GA)n and (GT)n sequence repeats. From 3000 initial selected colonies only 40% have displayed good size insert including microsatellite sequences. 871 of these clones have been sequenced. We present the results obtained for the first set of 91 couples of designed primers tested with 60 accessions of the Aurantioideae sub-family. 48 couples of these primers display clear mono focus mendelian segregating ratio. For these markers, amplifications were always successfull for true citrus (Poncirus, Fortunella and Citrus) Severinia buxifolia and Eremocitrus glauca the success of amplification was quite low for Clausenae tribe. From 10 accessions analysed at each level, the number of allele per locus changed from 1 to 9 (average: 3.45 allele per locus) at the intraspecific level (C. reticulata) and between 3 to 11 (average: 6.29) at the intrageneric level (Citrus; 10 species represented), Average observed heterozygosity are respectively 0.50 and 0.53 for each of them. These values are higher than ones observed for other codominant markers such as isozymes or RFLP. Neighbour-joining classification of true citrus is in accordance with the known relationship and evolution of citrus genera and species. Additionally we recommend a sub-set of markers as molecular descriptor for global Citrus germplasm management. (Texte intégral)