Two equations of state (PC-SAFT and CPA) are used to predict the monomer fraction of pure associating fluids. The models each require five pure-component parameters usually obtained by fitting to experimental liquid density and vapor pressure data. Here we also incorporate monomer fractions measured using spectroscopy, resulting in models that better predict the monomer fraction (fraction of molecules not participating in hydrogen bonding), without sacrificing the accuracy of the liquid density and vapor pressure correlations. Thus, it is clear that monomer fraction prediction depends on the way the parameters were obtained. The selection of appropriate association schemes is also investigated using spectroscopic data. For pure water a four-site scheme is shown to be the most appropriate scheme. In the case of pure alcohols, a three-site scheme is best for methanol; two-or three-site schemes perform about equally for ethanol; for higher alcohols a two-site scheme is preferred. This is in accordance with steric arguments. Some difficulties in the interpretation of spectroscopic data and their comparison with the predictions of association models are illustrated. Apparently anomalous data from different sources are shown to be consistent if interpreted correctly.