We describe an X/15 translocation which was balanced in a phenotypically normal mother [46,X,t(X;15)(p22;q15)] and unbalanced in her phenotypically abnormal daughter [46,X,der(X),t(X;15)(p22;q15)mat]. A third case involves a balanced X/21 translocation in a girl with a multiple congenital anomaly-retardation syndrome [46,X,t(X;21)(p11;p11?)]. 5-BrdU acridine orange banding on lymphocytes revealed late replication of the normal X chromosome in the mother and of the normal or abnormal X chromosome in the two other cases. Our findings are only partially consistent with previous observations. All X-inactivation patterns can be explained by random inactivation and subsequent selection against specific cell lines. Furthermore, the findings in our patient with X/21 translocation support the hypothesis of the existence of one inactivation center on Xq.