In a longitudinal field experiment, 42 groups were assembled on the factors of individual ability and attitude similarity resulting in a 2 x 2 crossed ANOVA design. These groups then worked on six projects which spanned a three-month time period. For each project, measures of group performance and group cohesion were taken. Analyses showed that high ability groups exhibited the best performance whereas attitudinally similar groups expressed the greatest cohesiveness. In addition, correlational data suggest that the effect of attitude similarity on cohension is not immediate, but develops over time, and that across the six projects, performance can be both positively and negatively related to cohension. (Author)