Aspects of processing raw analytical data to correct for a finite probe size and beam-specimen interactions are discussed and demonstrated for the case of interdiffusion in lattice-matched InP/(In, Ga)As heterostructures. At first, the probe size of the microscope was evaluated following the method of Michael and Williams. Then the two principal factors, affecting the results of composition profile measurement, were taken into account, namely the sample drift during prolonged data acquisition and the beam broadening by elastic scattering inside the foil. A method of quantification of the sample drift effect was proposed, as well as a method of final scaling of the composition profiles. Using this approach, we can get closer to measuring the true changes of composition profiles caused by interdiffusion at the nanometre level. (C) 2002 Elsevier Science B.V. All rights reserved.