This study aims mainly to clarify the effects of the multi-axial stress components on the initiation and growth behavior of creep voids. For this purpose, creep tests, particularly creep interrupt tests are conducted in tension, torsion and in combined tension-torsion stress states at 700°C, using tubular specimens of the austenitic steel SUS310S with high ductility. Creep voids formed in the specimens are examined in detail by observation with the scanning electron micrographs. It is found that creep voids in the torsional creep specimens form easily to a certain size in the primary creep, but they grow difficultly to rupture. While from torsional stress state to tensile one, creep voids become easy to grow. The initiation and growth behaviors of creep voids under the multi-axial stress conditions are discussed. It is further suggested that the von Mises equivalent stress is a dominant component for the initiation of creep voids, and the mean stress component strongly promotes their growth.KEY WORDS: creep; creep interrupt test; void; initiation and growth; multi-axial stress; fracture. creep fracture mechanisms, creep tests were conducted in tension, torsion and in combined tension-torsion stress states using the tubular specimens. In the combined tension-torsion stress states, the ratios of applied tensile stress s to applied shear stress t on the external surface of specimen are 1 : 1 and 2 : 1. These tests were performed in air using a tensile-torsional creep test machine. The test temperature was kept at 700Ϯ1°C and monitored with a thermoelectric couple attached on the external surface of each specimen. Tensile creep elongation and torsional creep rotation angle were measured continuously, using a dial gauge and a rotary encoder device respectively. Thus, the von Mises effective strain of the specimen at any time can be calculated by where e and g are the nominal tensile strain and the nominal shear strain, respectively. To investigate the effects of multi-axial stress components on the initiation and growth behavior of creep voids, 2 kinds of the creep interrupt tests for the specimens loaded at a given maximum principal stress s 1 were carried out: (1) The creep tests were interrupted at the initial and final stages of steady state creep. (2) The creep tests were interrupted as their effective strains reached the same.For the tested specimens, longitudinal sections were etched with the chemical solution (25vol%HNO 3 ϩ50vol% HClϩ25vol%Glycerin) after mechanical polishing. The specimens were then observed in detail with a scanning electron microscope (SEM). The "fraction of creep voids on grain boundary lines", 11) i.e., a fraction of total length of creep void areas on the grain boundary lines, was measured on the SEM micrographs. Also, the "average diameter of creep voids" was calculated. Because each void area observed on the longitudinal sections was just only a cutting plane of a spherical void, in this work the average diameter of the spherical creep voids was calculated approximately from these cut...