I. Hierarchical relationships in the blue gourami, T. trichopterus, are characterized by two distinct temporal phases — an ephemeral period of dominance establishment and a period of hierarchical maintenance. Variable behavior patterns are shown in the mutual fighting during dominance establishment between pairs of fish, but the initial outcome is always a discrete event, i.e., one fish dominates the other. The influence of several parameters on this period of dominance establishment has heen investigated in this study, and the probable relationships between these parameters and patterns of behavior characteristic of ultimate winners and losers has been discussed. 2. Prior residency per se does not increase the "dominance potential" of this species. Residency was not found to act as a positive input for increased aggressiveness; rather it was concluded that the "fright" component due to moving a fish into unfamiliar surroundings acts to inhibit the "normal" expression of aggressive behavior. The residency factor interacts both with prior dominance experience and with time since that experience. 3. Prior experience as a dominant increases the probability that a fish will win a subsequent encounter, while fish previously dominated within the last 24 hr seldom win subsequent encounters. Forty-three behavior measures recorded during these subsequent encounters did not indicate an arousal phenomenon due to dominating a partner for 24 hr, but a highly significant inhibitory effect on all agonistic behaviors was observed for those fish which had been dominated for 24 hr. These inhibitory effects wane during the next 24 hr, provided the dominating partner is removed. It is conceivable that short-term arousal input due to winning per se may temporarily lower aggressive thresholds, but whether these lowered levels would be "transferred" into a new fighting context is questionable. 4. The inhibitory effects of "non-residency" are in part dependent upon the "dominance state" of a fish. Dominant-experienced fish are only slightly inhibited by non-residency, while subordinate-experienced fish are strongly inhibited by this factor. A s the prior experience effect wanes, the fright due to unfamiliarity with surroundings is less inhibiting. 5. Initially, the decrement in responses due to prior experience is greater than the decrement due to non-residency. 6. Two measures of intra-individual sequence uncertainty (entropy) were calculated. The first included the uncertainty associated with all the elements defined for the system, i.e., [L, O, BS, F,, S, and P], while the second entropy values were based on only the agonistic set [L, O, BS, and F]. The prior experience factor influenced both entropy values more strongly than it did any other single variable. 7. The relative sizes of opponents seem paramount in deciding bout outcome when standard length differences (Δ S.L.) are beyond the range tested here, i.e., where Δ S.L. > 5 mm. As the time since the last dominance experience increases, the influence of Δ S.L. also increases. The length of time that two fish fight was shown to depend upon Δ S.L., while the only behavior correlated with relative size was BS rate. Smaller fish bit more frequently than did their counterparts. Finally, entropy measures were found to be functions of Δ S.L. Since entropy values increase as Δ S.L. approaches zero, entropy may contribute more to the "dominance vector" when fish are of nearly equal size. Other variables, such as biting intensity, may be more important as "dominance vector" components when one fish is considerably larger than its partner. 8. A "dominance vector" was defined in this study as that complex of variables which contribute to the defeat of one fish by another. Such a complex, highly integrated variable remains hypothetical. Some possible components of this vector indicated by this study include: variables associated with fin tugging rate or duration, biting intensity, unpredictability, and redundancy of certain inter-individual or intra-individual dyads such as [B1-Bw] or [LW-B1] where their conditional probabilities may be important. 9. A simple descriptive model relating environmental, experiential, size, temporal or other parameters to the dominance vector was presented. A mechanism describing how such a vector might operate in determining bout outcome was also postulated.