The current study aims at investigating the impact of homogeneous versus heterogeneous collaborative learning grouping in multicultural classes on the students' achievements and attitudes towards learning science. In the present study, heterogeneity was unpacked through two dimensions: the cultural background, represented by the different nationalities present in the class and the students' different abilities. The interaction between these two factors and their combined effect on the achievement and attitudes were also investigated. The study also considered an approach to provide quality teaching for a diverse group of students by neutralizing the heterogeneity factor or reducing its negative effect. For this purpose, 100 nine graders from more than 10 countries in an independent preparatory school in Doha / Qatar were divided into four classes and distributed over the following learning "STAD" groups: (1) Heterogeneous by ability but homogeneous by nationality; (2) Heterogeneous by nationality but homogeneous by ability; (3) Entirely heterogeneous (i.e. by both the ability and the nationality); and (4) Entirely homogeneous (i.e. by both the ability and the nationality). A diagnostic placement test, standardized pretest and posttest in addition to the regular school tests were used to measure the achievement of the students. A Questionnaire was developed to measure the attitudes of the students towards learning science as well as towards group working. The study concluded that the main effects of group structure on the students' attitudes towards learning science were demonstrated by the heterogeneous group. It affected all the attitude components except the "working with students from different cultural backgrounds" dimension, where nearly all group types had the same effect. However, this positive attitude was enhanced when the effect of mixed ability classes was combined with the effect of multiculturalism. Having foreign students or students from different cultures in a mixed ability class, yielded the best desired results. Therefore, the researcher highly recommends maximizing the heterogeneity in a class in all possible ways. The implemented collaborative learning strategy made learning more fun and beneficial for the students, enhanced their self confidence, academic awareness, and consequently their overall attitude towards science. The heterogeneity factor had a negative effect on the achievement of the students. The students in the mixed ability classes scored less than the students in other groups. However, when the "same ability" groups contained students from different cultural backgrounds, the results were the most favorable. The optimum class composition that may yield best achievement results and constructs positive attitudes is a compromise that maximizes group diversity and prevents individual isolation. The interaction between the two factors (ability + multiculturalism) gives the best desired results. In multicultural classes, collaborative learning should be supported by a multicultural education program, otherwise it would have little if any positive effect on the students achievement and attitudes towards learning science. (Contains 21 tables and 1 figure.) [Abstract modified to meet ERIC guidelines.]