A model based on probability theory is proposed to determine the orientation of the tectonic stress field from fault slip and earthquake focal mechanism data. On the assumptions that (1) slip and shear stress direction on the fault plane coincide, and (2) measured slip direction and fault plane orientation are random variables following the von Mises and the Fisher distribution, respectively, the effects of measurement errors in both variables on stress field inversion are considered. The distribution of shear stress direction on the fault plane can also be approximately modeled by the von Mises distribution, with the concentration parameter varying from infinity to zero according to the orientation of the fault plane, the direction of the principal stress axes, and the stress shape parameter. Measurement errors in both slip direction and fault plane orientation give rise to misfit angles between the predicted shear stress and the measured slip direction even if the tectonic stress field is uniform. A new inversion method is proposed that takes both measurement errors simultaneously into account. This method makes it possible to evaluate statistically the extent and significance of the deviations of the stress field from uniformity, i.e., the likelihood that a given fault population originates from a constant tectonic stress state. For sufficiently large sample size, misfit angles should belong to a von Mises family if the stress field is uniform. A significant deviation from a von Mises distribution implies that the tectonic stress field is not uniform. A few examples of application of the new inversion method are presented, together with a comparison with the results obtained from different methods and an assessment of the uncertainties in the inverted stress field. ¿ American Geophysical Union 1993