Given the peculiarities of mobile computer-supported collaborative learning (MCSCL) environments, forming suitable groups in such learning environments represents a hard and time-consuming task. This is because many conditions related to mobile learners, devices, and environment should be considered. Unlike the existing solutions, the present paper shows a grouping approach that allows a customizable formation of (1) homogeneous groups, (2) heterogeneous groups, and (3) mixed groups. The proposed solution does not only help instructors to dynamically form appropriate MCSCL groups, but it also allows to continually control the learners' learning, psychological, and social developments. To assess the effectiveness of the proposed solution, three metrics were used: (1) comparison between the characteristics of the existing group formation tools, (2) average intra-cluster distance of each grouping algorithm, and (3) an experimental evaluation in a real world environment. The obtained results show a great superiority of the proposed solution compared to the existing ones.