Fluorescence correlation spectroscopy (FCS) has become an important tool in polymer science. Among various other applications the method is often applied to measure the hydrodynamic radius and the degree of fluorescent labeling of polymers in dilute solutions. Here we show that such measurements can be strongly affected by the molar mass dispersity of the studied polymers and the way of labeling. As model systems we used polystyrene and poly(methyl methacrylate) synthesized by atom transfer radical polymerization or free-radical polymerization. Thus, the polymers were either end-labeled bearing one fluorophore per chain or side-labeled with a number of fluorophores per chain proportional to the degree of polymerization.The experimentally measured autocorrelation curves were fitted with a newly derived theoretical model that uses the Schulz-Zimm distribution function to describe the dispersity in the degree of polymerization. For end-labeled polymers having a molecular weight distribution close to Schulz-Zimm, the fits yield values of the number-average degree of polymerization and the polydispersity index similar to those obtained by reference gel permeation chromatography. However, for the side-labeled polymers such fitting becomes unstable, especially for highly polydisperse systems. Brownian dynamic simulations showed that the effect is due to a mutual dependence between the fit parameters, namely, the polydispersity index and the number-average molecular weight. As a consequence, an increase of the polydispersity index can be easily misinterpreted as an increase of the molecular weight when the FCS autocorrelation curves are fitted with a standard single component model, as commonly done in the community.