AbstractI. We recorded the responses of male Betta splendens when presented with various stimuli behind a transparent partition. The stimuli comprised i) live male and female B. splendens; ii) lifelike (motionless) models of male and female B. splendens; and iii) six stylised model B. splendens. The stylised models consisted of three pairs, one member of which had long male-like fins and raised opercula, and the other with short female-like fins and lowered opercula. The first pair had "aggressive", the second, "submissive", and the third, "reproductive" colouration. 2. The use of two techniques for comparing responses to the different stimuli was compared. Principal component analysis of unscaled data produced more satisfactory results than did clustering by single linkage on the basis of taxonomic distances calculated on scaled data. 3. The response by a male to a stimulus consisted of two categories of behaviour: "agonistic" display and nest-oriented activities. 4. The responses towards live males and females differed only in the relative amounts of "agonistic" and nest-oriented behaviour performed to each. A greater amount of "agonistic" behaviour was performed towards males. 5. Three main influences were found affecting a male's response to a conspecific: i) a tendency to perform all components of "agonistic" display and to inhibit performance of nest-oriented behaviour; ii) and iii) tendencies to perform particular components of the "agonistic" display. Thus the "agonistic" display was found to be complex and governed by at least three separate motivations. 6. The first influence varied according to the (perceived) sex of the stimulus, being great when the stimulus was a male, and slight when it was a female. The other two influences were independent of the nature of the stimulus. 7. Responses to lifelike, motionless models were very similar to those to live fish. 8. All male models and the "aggressive" female model were treated similarly to the live male. The other models were treated similarly to the live female. 9. A mechanism for sexual discrimination was proposed, in which a male is characterised by long fins and/or raised opercula, or, an unpatterned body. A female is characterised by short fins and a patterned body. Discrimination can be accomplished solely by use of visual stimuli.