The implementation for the first cycle involves knowing the end point of curve A, which was obtained originally by first shifting the steel envelope stress–strain curve to a point where elastic unloading results in zero stress, and then estimating the stress in the shifted curve at zero strain. However, Lacaze compared results obtained with this modeling approach and test results for the first cycle, and improved the response by taking the final point strain value equal (absolute) to the maximum attained in the opposite direction. For subsequent cycles the start and end points correspond to points (strains) at the previous unloading and reloading points. After reaching the end of curve A, it returns to the shifted monotonic envelope. For internal cycles the end point corresponds to the start of the outer loop, rather than the inside loop (Fig. 5). The need of a cyclic material lays in the fact that under monotonic compression of a bar and before instability clearly affects the response, all fibers are under compression. At larger compressive strains and considering that instability results in buckling the most compressive fiber goes further in compression (inside of concave zone of buckling), whereas other fibers towards the other end of the section would revert the strain direction resulting in unloading. In the case of cyclic analysis with buckling the entire material cyclic definition is required.