True triaxial tests ( σ 1 ≥ σ 2 ≥ σ 3 ) were conducted on the Zigong sandstone with a relatively low porosity of ~6.5% with the aim of assessing the influence of the intermediate principal stress σ 2 on the strength and deformability. Three series of true triaxial tests ( σ 3 = 0, 20, and 60 MPa) were performed. Within each series, σ 2 was planned to be varied from σ 2 = σ 3 to σ 2 = σ 1 from test to test. For each test, σ 1 was raised monotonically to failure while σ 2 and σ 3 were kept constant and the post-peak behavior has been captured. It is found that the strain in the σ 3 -direction is always larger than that in the σ 2 -direction under true triaxial stresses, which is attributed to the stress-induced tensile microcracks aligned with the σ 1 - σ 2 plane. For a constant σ 3 , the dilatant strain in the σ 2 -direction is progressively suppressed and becomes compactive as σ 2 raises from σ 2 = σ 3 to σ 2 = σ 1 . Based on a modified Hoek-Brown failure criterion with three stress invariants, the best fitting brittle strength envelope reveals that the σ 2 effect on rock strength, i.e., the ascending-then-descending trend for a given σ 3 , is the combined effect of mean stress and Lode angle. In the series of biaxial tests ( σ 3 =0 MPa), macro tensile fractures that parallel to the σ 1 - σ 2 plane develop regardless of σ 2 , displaying a failure plane angle of approximately 90°. As σ 3 is raised, failure plane angle generally decreases. When σ 3 = 20 MPa, the failure plane angle generally increases with σ 2 while there is no evident dependence on σ 2 when σ 3 = 60 MPa. Comparisons with the other two sandstones with higher porosity reveal that porosity exerts critical control on the brittle strength and that the σ 2 effect on deformability is also the result of two competing effects induced by mean stress and Lode angle. It is also found that the residual strength data facilitate the prediction of brittle-ductile transition pressure when σ 2 = σ 3 in spite of relatively low σ 3 level, which matches well with the empirical relation proposed by Mogi.1