Results of a traditional sedimentation technique for grain-size analysis (the sieve-pipette method) were compared with those of a laser diffraction grain-size analyser, the Coulter LS-100, Milled quartz samples and silty soil samples were used to investigate the effect of particle shape. The reproducibility of the Coulter LS-100 was better than that of the sieve-pipette method, except for the sand fraction. The agreement between the two methods is strongest for the milled quartz samples. The Coulter LS 100 underestimated the clay content of the silty soil samples and generally overestimated the clay content of the milled quartz samples. This indicates the importance of particle morphology. The milled quartz grains are very angular and somewhat elongated whereas the clay minerals in the clay fraction of the silty soil samples usually have a platy shape. Reduced major axis relationships are calculated which may be used to convert Coulter LS-100 results to those of the sieve-pipette method. Correlation is strongest for percentiles size data. The Coulter LS-100 underestimates the clay, but correctly estimates the and fraction. This results in a weak relationship for the total silt fraction as measured by the two techniques. Therefore, an alternative method is proposed for estimating this fraction. A comparison of our data with published data and relationships suggests that the calibration relationships are quite robust as long as the clay mineralogy of the samples is similar. (C) 1998 Elsevier Science B.V. All rights reserved.