X chromosome inactivation is unique among dosage compensation mechanisms in that the two X chromosomes in females are treated differently within the same cell; one X chromosome is stably silenced while the other remains active. It is widely believed that, when X inactivation is initiated, each cell makes a random choice of which X chromosome will be silenced. In mice, only one genetic locus, the X-linked X controlling element (X ce), is known to influence this choice, because animals that are heterozygous at X ce have X-inactivation patterns that differ markedly from a mean of 0.50. To document other genetic and epigenetic influences on choice, we have performed a population-based study of the effect of X ce genotype on X-inactivation patterns. In B 6 CAST F(1) females (X ce(b)/X ce(c)), the X-inactivation pattern followed a symmetric distribution with a mean of 0.29 (SD=0.08). Surprisingly, however, in a population of X ce(b)/X ce(c) heterozygous B 6 CAST F(2) females, we observed significant differences in both the mean (p=0.004) and variance (p=0.004) of the X-inactivation patterns. This finding is incompatible with a single-locus model and suggests that additional genetic factors also influence X chromosome choice. We show that both parent-of-origin and naturally occurring genetic variation at autosomal loci contribute to these differences. Taken together, these data reveal further genetic complexity in this epigenetic control pathway.