It is widely assumed that the vortex in a cyclone has a well defined length, which may be shorter than the physical length of the cyclone. One speaks of the ‘end’ to the vortex, and of the ‘natural vortex length’ or the ‘natural turning length’. The space below the vortex is normally assumed to be ineffective for separation, and the transport of solids along the wall is also inferior there. The axial position of the end of the vortex is thus an important design variable, yet little research work has been dedicated to its determination. In the course of research aimed at formulating a new relation for the natural vortex length, two experimental methods for determining this parameter have been found: (1) the position of the end of the vortex was visible for a period of time after smoke had been introduced in a glass cyclone; and (2) the position was found to be visible in the pattern of wall deposits after experimental runs with gas/solid cyclones. The paper presents results obtained in variations on a standard cyclone geometry with a tube section connecting the dust collection vessel to the cyclone. In these geometries the end of the vortex was in most cases located in the tube section. Comparisons are made between the two above methods of determining the length of the vortex, as well as comparisons with published correlations. The effects on the natural vortex length of the inlet gas velocity, the dimensions of the vortex finder and the length of the cyclone, and the solid loading of the cyclone charge gas are shown. A substantial effect of the length of the vortex on the separation efficiency of the cyclone is also demonstrated.