Abstract: Selectivity estimation of path expressions is the basis of XML query optimization and also intense research interest. A path expression may contain multiple branches with value predicates. Some of the values and the nodes of the XML data are highly correlated. Previous methods of selectivity estimation rarely take that relation into consideration, and assume, instead, that the selectivity of attribute values on different nodes and structures is independent and uniform. In this paper, a novel value histogram is proposed, which captures the correlation between the structures and the values in the XML data. Also defined are six operations on the value histograms as well as on the traditional histograms that capture nodes positional distribution. Based on such operations, the selectivity of any node (or branch) in a path expression can be estimated. Experimental results indicate that the method provides accuracy especially in cases where the distribution of the value or structure of the data exhibit a certain correlation without any independent assumption.