This paper presents a statistical analysis of about 50000 petrophysical data measured on core samples from the Continental Deep Drilling Project (KTB) of the Federal Republic of Germany. The scattering of the data must be taken into consideration using empirical relationships between pairs of parameters, e.g., sound velocity, density, heat production and thermal conductivity. Such covariances of parameters were calculated and used to find the principle components by applying factor analysis. The reduction of parameters by factor analysis may help other scientists to concentrate on the essential parameters to be measured. About 50% of the variance of all data can be explained by one background variable, the so-called lithology factor. The variables that load a factor are either highly correlated or anticorrelated. The lithology factor combines gamma spectroscopy data, density and the mineral contents of quartz, amphibole, garnet and white mica. Seismic velocities and porosity data, however, were less well related to the lithology factor. Therefore, the KTB data indicate that correlation between seismic velocity and one of the lithology factor loading variables is unlikely. The lithology factor distinguishes 3 major rock types: metabasites, gneisses and an intermediate type. The variance of the petrophysical parameters within the rock types, plotted in crossplots, show the level of validity of commonly used relationships among these parameters. Measurements under ambient and simulated in-situ conditions are included to enable discussion of chemical, mineralogical and microstructural characteristics of the rocks.¿ 1997 American Geophysical Union