A technique is described for measuring load magnitude and resultant load contact location in the facet joint in response to applied loads and moments, and the technique applied to the canine lumbar spine motion segment. Due to the cantilever beam geometry of the cranial articular process, facet joint loads result in surface strains on the lateral aspect of the cranial articular process. Strains were quantified by four strain gages cemented to the bony surface of the process. Strain measured at any one gage depended on the loading site on the articular surface of the caudal facet and on the magnitude of the facet load. Determination of facet loads during in vitro motion segment testing required calibration of the strains to known loads of various magnitudes applied to multiple sites on the caudal facet. The technique is described in detail, including placement of the strain gages. There is good repeatability of strains to applied facet loads and the strains appear independent of load distribution area. Error in the technique depends on the location of the applied facet loads, but is only significant in nonphysiologic locations. The technique was validated by two independent methods in axial torsion. Application of the technique to five in vitro canine L2.3 motion segments testing resulted in facet loads (in newtons, N) of 74 + /-23 N (mean +/-STD) in 2 newton-meter, Nm, extension, to unloaded inflexion. Lateral bending resulted in loads in the right facet of40 + /- 32 N for 1 Nm right lateral bending and 54 +/ - 29 N for 1 Nm left lateral bending. 4 Nm Torsion with and without 100 N axial compression resulted in facet loads of 92 + /-27N and 69 + /-19 N, respectively. The technique is applicable to dynamic and in vivo studies.