Collaborative Virtual Environments (CVEs) are distributed virtual worlds where users can interact simultaneously with synthetic objects to perform task. Coordination and awareness are the complementary elements for collaboration which can be achieved through e cient communication. Loose coordination, no task distribution strategy, network latency, less awareness and low communication among collaborating users are some of the main issues that degrade users performance in CVE. In addition, collaborative tasks without using any proper cognitive aids are very di cult to perform. In this research study, a novel task distribution mechanism has been proposed for e cient collaboration. The mechanism de nes task distribution strategy (i.e. statically or dynamically) among users in CVEs. In static task distribution there exist loose dependency among task/sub tasks and requires less communication for execution while in dynamic distribution users are more dependent on each other and thus requires more communication. The main objective of proposed mechanism is to manage and control the concurrent users actions for better performance. The basic theme of this thesis is to increase users performance in CVEs. In this context, we have developed di erent CVEs and conducted experiments. The experiments were carried out in research lab in University of Malakand. Users were asked to perform the assembly of the word "UNIVERSITY" from synthetic objects lying in di erent rooms in VE for checking the users performance and four navigation guidance technique comparisons. Also a chemistry lab was developed to conduct experiment to check user performance and learning capability. Analysis was done on the basis of task completion time, errors and users'' learning. Results of the experiments revealed that whenever the user performance is the main motives for executing the task in CVEs, static task distribution mechanism should be adopted and whenever learning is the main objective then dynamic task distribution is preferred. Also to evaluate the impact of 3D interaction guidance metaphors on user performance in a CVEs an experiment was conducted with 3D Map with Line (3DML), audio, textual and arrows guiding techniques for navigation. The results showed that arrows guide in CVEs improve users performance, then 3DML, textual and audio guiding techniques respectively. Thendings of this research work contribute to the development of a collaborative systems for tele-operation, assembly task, e-learning, computer aided design and entertainment.