The geochemical logging tool (GLT) provides in situ measurement of the concentrations of 10 elements commonly found in sedimentary rocks: Si, Al, Fe, Ca, K, Ti, S, Th, U, and Gd. The element abundances thus determined can be used to derive lithology, sand class, mineralogy, porosity, permeability and cation-exchange capacity. The usefulness of this tool will depend upon the precision and bias of measurements. This paper describes the results of deploying a GLT at the Imperial College borehole test site in a sequence of finely layered siliciclastic and carbonate rocks. The bias of the GLT was quantified for six elements (Si, Al, Fe, Ca, K and Ti) by comparison with measurements on corresponding core samples made by inductively coupled plasma atomic emission spectrometry (ICP-AES). The high density of core measurements (approximately every 25 cm) allowed them to be averaged over a depth interval equivalent to the vertical resolution of the GLT thus reducing errors due to differences in sample size. A statistically significant bias for the GLT for all elements was calculated from a simple linear regression; values were between ^–55 and +5%. The detection limits and dependence of precision on concentration for Si, Al, Fe, Ca, K and S were determined using duplicate analyses from two logging runs in the same borehole. Values for the detection limits ranged from 0.26 for Fe to 2.70 mass-% for Si. When the precision varied with concentration, the high level precision was found to be less than 6.5% in most cases. It was found that the processing applied to the raw counts data obtained from the GLT played a major role in the determination of detection limits, precision and also the accuracy of measurements obtained using the GLT.