Assuming the signal response in a pulsed gradient stimulated echo (PGSTE) experiment (on a polymer) to be described by a simple stretched-exponential function (SEF) and knowing the scaling law between diffusivity and molecular weight, the molecular weight distribution (MWD) characteristics (kurtosis, skewness, moment, and width) are derived and compared to the corresponding distribution characteristics obtained by a log-normal function fi t (to the same MWD). Also, the challenge involved in obtaining a reliable weight average molecular weight from an SEF response function is discussed.This paper was amended because of mistakes in the equations 11, A3, and A6. polymers in solution can be well approximated by a stretched exponential function (SEF). [ 6 ] Because of the inherent nature of the PGSTE NMR experiment, the signal response from a polymer dissolved in a solvent can be represented by the Laplace transform (LT) of a distribution function, namely, the distribution of diffusivities. Or, alternatively, the distribution function can be expressed by the inverse Laplace transform (ILT) of the response function. It is well known, however, that this is an illposed numerical problem of particular complexity. However, it is known that the ILT of an SEF can be expressed explicitly by a series expansion, [ 7 ] which then bypass the numerical problem of performing an ILT of a SEF.We recently presented an alternative approach to overcome the above numerical problem [ 8 ] whereby a numerical Laplace transform of a log-normal distribution (LND) function was performed and the resulting transform was fi tted to an SEF. By applying the well-known relation between the molecular weight and molecular diffusivity, two empirical equations relating the two SEF parameters of the PGSTE NMR response function and the two corresponding parameters of the log-normal molecular weight distribution (LNMWD) were derived. The above method is, by nature, somewhat approximate as it implicitly