Dry and wet brushite cements with various solid/liquid ratios were tested in compression and tension. Two different testing techniques were used to determine tensile strength: Direct Tensile test (DT) and Diametral Compression test (DC) (Brazilian test), which is an indirect way of measuring tensile strength on brittle materials. Statistical analysis of the results obtained on dry cements points out a constant ratio between the values measured by DT and Brazilian tests (DC/DT = 85%). The Mohr's circles representation allows us to understand that, for a material like our cement, ultimate stress measured with the Brazilian test can only underestimate tensile strength, because the compressive/tensile strength ratio is lower than 8. The second consequence of this low ratio is that, in the Brazilian test, the plane along which fracture initiates undergoes not only a normal tensile stress, but also a tangential stress component. Thus, the state of stress on the fracture plane differs from the one taking place in the direct tensile test. Consequently, with such a material (sigma(c)/sigma(t) < 8), the Brazilian test does not estimate the true tensile strength.