The complex moduli of a series of polypropylene melts were measured together with their gel permeation chromatograms. A model, as proposed by one of the authors (A. Schausberger), was used to correlate the molar mass distribution data with the dynamic mechanical data. It turns out that, in particular, the storage modulus is so sensitive to variations in the high molar mass tail of the distribution that gel permeation chromatography (GPC) is hardly a method which can be used to check the validity of molecular theories on polymer melt rheology. It seems that a total mass fraction of less than a half percent of large molecules is decisive for almost the whole shape of the modulus versus frequency curve. Only a proper extrapolation of the molar mass distribution into the high molar mass region (assumption of a lognormal tail) leads to a favorable comparison between the shapes of predicted and observed frequency dependences.