THIS STUDY TESTS THE HYPOTHESIS THAT BIRACIAL GROUPS IN WHICH RACEMATES WERE PAIRED FOR DIRECT CONTACT WOULD PERFORM BETTER THAN THOSE IN WHICH RACEMATES WERE ISOLATED. IT ALSO ASSUMED THAT ETHNOCENTRIC SUBJECTS WOULD PERFORM BETTER WITH CONTACT WITH A RACEMATE, AND THAT AGGRESSIVE SUBJECTS WOULD GET ALONG WELL WITHOUT IT. THE SUBJECTS WERE BOYS FROM A MULTIRACIAL HIGH SCHOOL IN A LOW SOCIO-ECONOMIC AREA OF DETROIT. EACH BOY DECIDED WHICH THING, FROM A CHOICE OF FOUR, ALL THE OTHER PARTICIPANTS (IDENTIFIED BY A COLOR PHOTOGRAPH) WHOM HE CONTACTED BY TELEPHONE, HAD IN COMMON. RESULTS SHOW THAT GROUP STRUCTURE MADE NO DIFFERENCE IN PERFORMANCE. THE AGGRESSIVE-XENOPHILIC WAS THE ONLY TYPE THAT DID IMPROVE WITH CONTACT BETWEEN RACEMATES. THE STUDY SHOWED THAT ALTHOUGH CONTACT BETWEEN RACEMATES INFLUENCES COLLECTIVE BEHAVIOR, ETHNOCENTRISM, AGGRESSIVENESS, AND RACIAL BALANCE DETERMINE WHETHER INTERACTION WILL FACILITATE OR INTERFERE WITH GROUP PRODUCTIVITY. THIS PAPER WAS PRESENTED AT THE AMERICAN PHYCHOLOGICAL ASSOCIATION ANNUAL MEETING (WASHINGTON, SEPTEMBER 1967). (SR)