Designers of context-sensitive program analyses need to take special care of the memory consumption of the analysis results. In general, they need to sacrifice accuracy to cope with restricted memory resources. We introduce χ-terms as a general data structure to capture and manipulate context-sensitivity analysis results. A χ-term is a compact representation of arbitrary forward program analysis distinguishing the effects of different control-flow paths. While χ-terms can be represented by trees, we propose a memory efficient representation generalizing ordered binary decision diagrams (OBDDs).