Annealed ultra-shallow implants. often develop sharp features in the top few nanometres of the wafer during thermal processing. In a SIMS depth profile, much of the feature can fall within the surface transient making quantification and dosimetry inaccurate. In this paper, we examine the errors involved for an annealed B implant in a silicon matrix with an exceptionally high surface concentration. Data from capped and uncapped samples is compared to distinguish artefacts of the measuring process from, real features. Cascade dilution was used to increase the accuracy of the dosimetry and it was found that in the uncapped samples, the dose is under-estimated for primary beam energies above 500 eV by increasing transient width and over-estimated at lower energies because of the matrix effect. We also discuss energy sequencing-obtaining profiles using a range of beam energies and extrapolating the shape to zero beam energy. (C) 2002 Elsevier Science B.V. All rights reserved.