Uniparental disomy (UPD) in humans is caused primarily by meiotic nondisjunction events, followed by trisomy or monosomy 'rescue'. The majority of cases appear to be associated with advanced maternal age, and may be initially detected as mosaic trisomies during routine prenatal diagnosis by chorionic villus sampling or amniocentesis. In addition, structural abnormalities including Robertsonian translocations, reciprocal translocations and supernumerary marker chromosomes appear to be associated with an increased risk of UPD. Predicting the phenotypic effects of UPD is complex, as three independent factors are involved: (i) effects of trisomy on the placenta or the fetus; (ii) autosomal recessive disease due to reduction to homozygosity; and (iii) imprinted gene effects for some chromosomes. To date, UPD in humans has been reported for 25 of the 47 possible uniparental types. Imprinting effects have been established with certainty for four human chromosomes that have homology to mouse chromosomes which have been shown to have significant phenotypic effects in uniparental animals. A normal phenotype has been reported for 14 other UPD types. Thus, collection of data on UPD cases in humans is providing an imprinting map analogous to the experimentally derived imprinting map in mouse. This human imprinting map has important clinical implications, particularly in the area of prenatal diagnosis.