A general theory is presented for computing the distributions and averages of the lengths of subchains and chain ends in cross-linked polymer networks. This theory was tested by computer simulations, and we show the results of both theory and simulations, for a better comparison. We find a significant difference between the values of chain ends and subchains depending on the extent of the reaction. As a consequence, the fraction of elastically active or effective material was clearly overestimated in previous works. We show how the different length distributions of intramolecular loops or cycles, subchains, and chain ends influence each other and how they can be combined into one general distribution. The influence of the spatial density or homogeneity of the chains, the functionality of the cross-links, the length distribution of the initial polymer chains, and other deviations from an idealized system are discussed briefly.