Double torsion testing can produce fracture toughness values without crack length measurement that are comparable to those measured via standardized techniques such as the chevron-notch, surface-crack-in-flexure and precracked beam if the appropriate geometry is employed, and the material does not exhibit increasing crack growth resistance. Results to date indicate that 8 < W/d < 80 and L/W > 2 are required if crack length is not considered in stress intensity calculations. At L/W = 2, the normalized crack length should be 0.35 < a/L < 0.65; whereas for L/W = 3, 0.2 < a/L < 0.75 is acceptable. In addition, the load-points need to roll to reduce friction. For an alumina exhibiting increasing crack growth resistance, values corresponding to the plateau of the R-curve were measured. For very thin plates (W/d > 80) nonlinear effects were encountered.