Multi-site MRI studies collect large amounts of data in a short time frame. Large sample sizes are desirable to address power and replicability issues that have been problematic for scientists in the past. Although multi-site MRI solves the sample size problem, it brings with it a new set of challenges. Scanning the same person at different sites might result in differences in MRI derived measurements. In this thesis we compared three approaches to facilitate the analysis of multi-site MRI data: quantitative R1 mapping, adding site as a covariate in a linear model, and using the ComBat method. We also investigated the relationship between two common MRI measurements: signal and volume. We collected data from 64 healthy participants across 3 GE scanners and 1 Siemens scanner at 3T. We found that signal intensity was different between vendors whereas volume was not. Our R1 method resulted in values that were different across vendor and significantly lower than those reported in the literature. B1+ maps used to calculate R1 were different across sites. Using a scale factor, we were able to compensate for mistakes in R1 mapping. We also found that adding site as a covariate corrected mean differences in signal intensity across sites, but not differences in variance. The ComBat method gave best similarity between sites. However, since different people were scanned at each site, we couldn’t evaluate the effectiveness of each method as variation in the data could have been due to site effects or heterogeneity in participants. White matter volume and signal intensity in the white matter were correlated in males but not in females. We found that this low correlation was caused by outliers in our female sample. The correlation between white matter volume and signal in males suggests that both metrics are measuring myelin and can be used as converging evidence to detect changes in brain myelination. / Thesis / Master of Science (MSc)