Male three-spined sticklebacks were held individually in compartments under one of four conditions: Total Isolation (Ti) ; Physical Isolation (Pi), compartments from which the males could see the room; Physical Isolation with Aggression Tests (PiA), similar to Pi but with the addition of aggression tests each day; and Physical Isolation with a Goldfish Companion (PiG), similar to Pi but with a white color-phase goldfish in the same compartment. Ti and PiA individuals became shy and tended to hide motionless in the corners of their compartments; when disturbed, they sometimes dashed madly about the compartment before becoming motionless. Pi and PiG fish became tame or non-shy. When tested individually in a 300 cm tank planted with a dense row of vegetation along one end, the shy males (Ti and PiA) selected nest sites nearer the ends of the tank than the non-shy males (Pi and PiG) did. The distribution of nests built by Ti individuals differed significantly from those of the other groups. Latency of response to a test male showed great variation; the difference between PiG with the shortest delays, and PiA with the longest delays, are significant. PiG and Pi males showed significantly more aggression toward the test male than did Ti and PiA males. The small differences in the amount of aggression between PiG and Pi and between Ti and PiA are not significant. Latency of response is negatively correlated with number of bites in the first minute and (less well) with total number of bites. The negative correlation between latency and total aggression is much better if we exclude those with short latencies. There is no correlation between nest site and latency of response to the test male. The location of the nest (distance from nearest end) and amount of biting directed at the test male are positively correlated. It is quite clear that the conditions under which these wild caught, first-year males were held in the laboratory immediately prior to the experiments affected the numbers nesting, their choice of nest sites, their latency in responding to intruding male, and the amount of biting they directed against them. The differences in treatment did not affect the percent of individuals that built nests and responded to the test male. Four Ti individuals showed large amounts of nest building behavior during the aggression tests, but none of the other groups showed this. In spite of gross differences in the treatments the four groups received, the two groups which became shy responded very similarly to one another and the two groups which became non-shy also responded similarly to one another.