The Midas cichlid (Cichlasoma citrincllum) is an abundant fish in the lakes of Nicaragua. Many populations are polychromatic, about 7 to 10% of the adults being variously white, yellow, orange, or red, and without the species-typical markings. These are termed 'golds' because the most frequent color is yellow-orange. The common cryptically colored morphs are called 'normals'. In the experiment, 6 equal sized small fish had to approach a feeder guarded by a Midas cichlid twice their mean weight. In the experimental groups, 3 of the small Midas cichlids were gold and 3 were normal (mixed color groups). The control groups were of two types, one with all 6 small fish gold, the other with all 6 normal (pure-color groups). In half of all trials, the large fish was gold, and in the other half normal in color. Data were gathered on Days I and 3. I) Differences in the behavior of large fish toward the small ones were not statistically significant. However, the large gold attacked the small fish more often than did the large normal, although the rate of attacking was remarkably low. The large gold fish also fed on average much more often (9.7/hr.) than did the large normal (0.7/hr.), whose mean rate was about half that for all small fish. 2) There was little difference in rate of feeding among the small fish. However, on Day 3, only, the small fish fed significantly less in the presence of a large fish whose color they shared. 3) The spread in weights within each group of small fish increased 1.4 to 2.4-fold during the four days of the experiment (growth depensation). 4) The distribution of the fish within the arena was recorded at 30-sec. intervals, and analyzed with regard to Near (to the feeder), Mid, and Far Regions, adjusting the data to fish per m2 per hr. There were no significant differences in distribution in relation to fish coloration. The small fish occurred at about the same rate in the Near and Mid Regions, but more frequently in the Far. The shelters in the Far Region had an effect equivalent to adding more space. 5) Across all experiments and groups of small fish, the rate of attacking was, in the Near, Mid, and Far Regions, respectively, 9.1, 19.4, and 14.7 per m2 per hr. When these data were modified to take into consideration the fish available for attack in the same region, the scores became 10.5, 22.4, and 15.0 per m2 per hr., respectively. Small fish, therefore, were about half as likely to attack in the Near Region, with the feeder and the large fish, as in the Mid Region. 6) The small fish attacked significantly less those small fish that shared the color of the large fish. 7) Small golds, in mixed color groups, were attacked less than were small normals. 8) Attacks by golds on normals were compared to those on golds by region and by day. Likewise, the attacks by normals on golds were compared to their attacks on fellow normals. In total, there were 24 such pairs of comparisons for the mixed color groups. Of these, 50% were significantly different. In 11 of these 12 cases, golds were attacked less than were normals. And 9 of these 11 cases had the large fish gold. Thus sharing the color of the large fish conferred some immunity from attack. When both the small and large fish were gold the effects summated in the favor of small golds to produce significant differences. Conversely, when the large fish was normal the effects of gold coloration and that of not sharing the color of the large fish cancelled one another. 9) All the small golds had had prior experience with normal colored siblings, but most of the small normals had never been with golds before. The attack suppressing effect of gold color on normals was pronounced on Day I but weak on Day 3. This suggests that the effect of gold coloration is enhanced by infrequent exposure. If so, golds should experience a lessened advantage in aggressive interactions in direct proportion to their abundance in a population. 10) We hypothesize that gold coloration decreases the readiness to attack in the perceiver by stimulating incompatible fear responses.