Abstract It is shown that the procedure of fitting two Lorentzian lines to an experimental Mössbauer spectrum consisting of a quardupole doublet of overlapping resonances leads to error in the measurement of line positions, and hence of quadrupole splitting, if the absorber is not thin. Two basic methods have been used to obtain accurate values of quadrupole splitting from thick absorber Mössbauer spectra. The first, calculation of correction factors to the results of Lorentzian fitting the thick absorber spectra, has itself been carried out in two ways. 1. 1. Thick absorber spectra have been simulated by computer and fitted by Lorentzian lines. Families of curves of have been plotted, using which the results of an experiment can be corrected assuming that the thin absorber linewidth is known. 2. 2. A simple formula for correcting thick absorber data has been derived, based on the assumption that a thick absorber spectrum approximates to the sum of two Lorentzian lines, but with parameters differing from those of a thin absorber spectrum. In the second basic method, a thick absorber lineshape function is fitted directly to the experimental spectrum. In addition to values of quadrupole splitting, this method is shown to give an estimate of the recoil free fraction of Mössbauer nuclei in the absorber. The methods have been applied to thick absorber quadrupole spectra of 119mSn in Y 2Sn 2O 7 and give results for quadrupole splitting which are in good agreement with a thin absorber measurement. A value for the recoil-free-fraction of 119mSn in Y 2Sn 2O 7 is also obtained.