Description: XCTD (eXpendable Conductivity Temperature Depth) probes, developed recently by SIPPICAN Inc., have been used simultaneously with a CTD sonde in order to test, in the field, their performance and accuracy (interpreted as ±2 standard deviations of the XCTD-CTD differences). We have taken advantage, during the THETIS-I experiment in March 1992, of both the homogeneous and the stratified areas encountered in winter in the northern part of the western Mediterranean Sea to differentiate the errors due to the experimental conditions from those effectively due to the sensors. Although some intrinsic problems are evident, so that only seven out of the nine probes considered for comparison are usable, the accuracy specified by the manufacturer for the temperature (AT = ± 0.03°C) is reached after standard processing, while the accuracies in conductivity, salinity and potential density are AC ≈ ± 0.06 mS/cm (the specified value is AC = ± 0.03 mS/cm), AS ≈ ± 0.04 and Aσθ ≈ ±3 kg/m3. However, when the experimental errors (in situ natural variability, relatively rough estimation of the XCTD depth) are considered, it appears that the effective accuracies of the XCTD sensors are better than ± 0.02°C and ± 0.04 mS/cm, that is to say better than and close to the specified values of ± 0.03°C and ± 0.03 mS/cm. Occasional offsets in conductivity can further be well corrected for by using a temperature-salinity relation in some limited depth range and area where this relation is known to hold well; the conductivity-sensor accuracy then significantly improves to AC≈ ± 0.02 mS/cm resulting, for our study area, in corresponding salinity and potential density accuracies of AS≈ ± 0.03 and Aσθ ≈ ± 0.02 kg/m3. Thus, such instruments promise to be useful tools for many experimental studies. Complementary comparisons, performed with new versions of the XCTD probes under less convenient experimental conditions, are also presented