This paper summarizes results of our investigations on: optimization of conditional sampling technique for eduction of naturally occurring large-scale structures in an axisymmetric mixing layer, comparison of the natural structure with that induced via controlled excitation, and the sensitivity of the educed structure to the excitation amplitude and of the natural coherent structure to Reynolds number and initial condition. Measurements include sectional-plane contours of various structure properties; however, coherent vorticity is the principal measure used in these considerations. All plausible alternative triggering criteria, based on reference velocity signals from fixed probes, were considered in order to arrive at the best practical eduction technique. It is shown that the simple criterion of triggering on the positive peaks of the longitudinal velocity signal derived from the high-speed edge of the mixing layer results in the optimum eduction. The characteristics of the natural structures, educed by the optimum detection criterion, are found to be independent of ReD over the measurement range 5.5 × 104−8 × 105. A mild dependence on the initial condition (viz laminar vs. turbulent) is observed, the structures being more disorganized for the initially laminar boundary-layer case. The educed natural structures agree well with those induced by controlled sinusoidal excitation at low excitation levels; higher levels, however, produce considerably stronger structures.