The finding of a supernumerary marker chromosome in amniotic fluid cells poses a considerable counseling dilemma. In 6,500 cases referred to our laboratory over a 4 1/2-year period, eight such cases were identified (0.123% of all cases). In five of the eight cases, a diagnosis of true mosaicism between cells with 46 and 47 chromosomes was made. In the remaining three cases, the marker was present in 100% of the cells. In three cases, the marker was determined to be familial in nature with mosaicism present in the parents of two of these cases. Detailed cytogenetic findings for each case are provided. In no cases were abnormalities noted in either abortuses or live borns. The high incidence of mosaicism in these cases seems to indicate a propensity for supernumerary chromosomes to be lost. Familial markers may not be passed on for many generations, and they may arise as new mutations relatively frequently. There is an urgent need for more information on the risks associated with the prenatal detection of supernumerary chromosomes. We recommend that in considering the implications of the prenatal detection of marker chromosomes cases be considered in at least four distinct groups: type 1--familial and nonmosaic; type 2--familial with mosaicism in either the amniotic fluid cells, a parent, or both; type 3--de novo markers and nonmosaic; and type 4--de novo with mosaicism present in the amniotic fluid cells.