The internal associations of comb-shaped associative thickeners are modeled by Gaussian statistics to predict the probability of chain collapse and that for various types of internal associations. In addition to the idealization of the polymer backbone to a random walk, the hydrophobic groups at the ends of each tooth of the comb are treated as point“stickers.” The model considers all the ways by which stickers can form clusters of two or more teeth. The number of teeth per chain is not limited by the model, but rather by computation facilities. It is predicted that even with weak pairwise sticker interactions (2kT) adding short teeth to a long backbone drastically increases the probability of some internal associations, though the probability of complete chain collapse decreases significantly as teeth are added, due to the effect of backbone entropy.