Twenty X-chromosomal short tandem repeat (STR) loci were typed in 80 families of Italian descent, composed by mother and two or more sons, for a total of 93 meiosis. The analyzed X-STR panel included six clusters of closely linked markers (each spanning less than 3cM): DXS10135-DXS10148-DXS8378 (Xp22); DXS7132-DXS10074-DXS10079 (Xq12); DXS6801-DXS6809-DXS6789 (Xq21); DXS7424-DXS101 (Xq22); DXS10103-HPRTB-DXS10101 (Xq26); DXS8377-DXS10134-DXS7423-DXS10146 (Xq28). Recombination fractions between pairs of markers calculated by pedigree analysis were compared with those obtained from the second-generation Rutgers combined linkage-physical map of the human genome. The observed differences confirm that recombination is not homogeneous along the X chromosome and that the conventional subdivision of X-STRs in four groups of completely unlinked markers cannot be regarded as true. Significant linkage disequilibrium was found between markers DXS6801 and DXS6809 (p=0.017). The effect on likelihood calculations of inferring haplotype frequencies from allele distributions rather than haplotype count in the relevant population was evaluated. (Published Abstract)