A reference population is used when integrating the individual components of the Multiple Sclerosis Functional Composite (MSFC) into a single composite score. The choice of reference populations may have a significant impact on the resulting MSFC score, yet the impact of different reference populations has not been evaluated. We evaluated the impact of different reference populations when deriving the Multiple Sclerosis Functional Composite (MSFC) in a group of MS patients followed longitudinally for two years. Reference populations included the study population at baseline (n = 60), a group of healthy controls (n = 18) and the National MS Society Task Force reference population (n = variable). We found that the choice of reference population had a significant impact on the resulting MSFC Z-score, sometimes compromising the statistical sensitivity to change over time. Our results suggest that longitudinal studies employing a multisystem composite Z-score should use a reference population with similar patients, which can most easily be achieved by using the baseline measures of the population under study. These results have significant implications to sample size estimates for longitudinal clinical studies and therapeutic trials.