In project planning, it is presumed that resources will be available to pick up their allocated tasks as planned while what happens in reality is rarely this simple. This discrepancy between planning and reality intensifies in Critical Chain Project Management (CCPM) portfolios where no level of multitasking is allowed. Regarding this, Ghaffari and Emsley (2016) determined the boundary between good and bad multitasking in CCPM portfolios with various resource capacities that showed some limited levels of multitasking could be beneficial to alleviate resource availability issues in such environments. In this paper, the authors aim to investigate how good and bad multitasking affect buffer requirements of CCPM portfolios with the same resource capacities considered in the above study. In a deductive approach, a hypothesis is developed and tested through experiments of ten portfolios with similar size and complexity levels, each one containing four projects with a different resource capacity, and comparing the results to their simulated counterparts obtained by the mentioned study. The results show that buffer requirements of portfolios with resource capacity of 130% and lower can be reduced through allowance of higher levels of multitasking. As a major contribution, a framework for buffer requirements of CCPM portfolios with different levels of multitasking and resource capacity is recommended.