Introduction: A series of studies have shown that circulating testosterone increases the persistence with which animals search for and select a particular visual cue to respond to (as a consequence of experience with it) - and hence incur a perseveration of response towards such cues (e.g. Rogers 1971; Andrew 1972, Archer 1976; Earley & Leonard 1978). We tested the effect of testosterone on such attention-related mechanisms by studying the effect of intradimensionsional colour changes to cues relevant and irrelevant to an operant discrimination on a continuous reinforcement schedule (non-reversal shifts). Methods: Treatment: Chicks (20 males) were given 12.5 mg testosterone enanthate im after training on a continuous reinforcement schedule of response (CRF); 19 controls received sesame oil vehicle and 8 males and 15 females had no treatment. Testing: Birds were given A) test sessions on day 10 and 11: whereby each session consisted of 2 minutes training, 5 minutes test then 2 minutes re-training: B) 4 types of test --(i) the negative key colour changed from red to deep blue, (ii) the positive key colour changed from pale blue to green, and (iii) both key colours changed as described, and (iv) the overhead lighting changed with the introduction of a pale red filter. Results: 1/ Treatment did not affect CRF patterns of discriminative responding. 2/ All birds decreased their response rate after a colour change. 3/ After changes on the non-reinforced key, testosterone treated birds showed significantly less attenuation of response rate. 4/ Testosterone treatment also maintained a higher rate of response (and hence fewer reinforcements) - seen especially after a change in the negative cue or overhead lighting vs. changes in the positive cue (irrelevant changes of stimulation). 5/ Testosterone treated birds also showed a shorter latency to respond after a colour change. 6/ The female birds did not differ from the males Conclusions: The results support a role for testosterone in the discrimination between relevant and irrelevant stimulus changes and the persistent expression of learned sets. The fact that responding after testosterone treatment was altered by irrelevant rather than relevant stimulus changes suggests that testosterone achieves its persistence effect by enhancing the activated set or what is at the focus of attention, rather than the inhibition of features irrelevant to the ongoing situation - these provide the reference for 'what is relevant' and changes in them disturb this mechanism.