The elevated X-maze has strong claims to validity as an animal model of anxiety, both in theoretical basis and drug responses. The model is sensitive to actual and putative anxiolytics, but because insufficient time has elapsed since its discovery, no agent first predicted to be anxiolytic in the elevated X-maze has been brought into general use yet. It has an advantage in detecting anxiolytic and anxiogenic agents under the same operating conditions. The design and execution of experiments with the model is discussed and it is shown that baseline arm preference and the size or direction of drug effects differ in the procedural factors affecting them. Because it presents features of both passive and active avoidance and approach/avoidance conflict, it may prove able to detect drug effects in different forms of human anxiety and ail in understanding their neurobiology.