A variable that leaves the basis can re-enter in a later simplex iteration. The simplex method explores adjacent extreme points, and a variable may become favorable to re-enter the basis if its reduced cost becomes attractive again during optimization.