Patients (n = 8) with uncomplicated mild traumatic brain injury (mTBI) underwent serial assessments (4) with diffusion tensor imaging (DTI) and neuropsychological testing within the first 8 days post-injury. Using a multi-case study design, we examined changes in brain parenchyma (via DTI-derived fractional anisotropy [FA], apparent diffusion coefficient [ADC], axial diffusivity [AD] and radial diffusivity [RD] in the left cingulum bundle) and in memory performance (via Hopkins Verbal Learning Test-Revised). Qualitative inspection of the results indicated that memory performance was transiently affected in most participants over the course of the week, with performance most negatively impacted on the second assessment (days 3-4 or 97-144 h post-injury), and then returning to within normal limits by 8 days post-injury. Alternatively, FA and other DTI metrics showed a more complex pattern, with the trajectory of some participants changing more prominently than others. For example, FA transiently increased in some participants over the study period, but the pattern was heterogeneous. Memory performance appeared to mirror changes in FA in certain cases, supporting a pathophysiological basis to memory impairment following mTBI. However, the pattern and the degree of symmetry between FA and memory performance was complex and did not always correspond. Serial imaging over the semi-acute recovery period may be important in reconciling conflicting findings in mTBI utilizing memory and/or DTI. Serial use of imaging modalities including DTI may aid understanding of underlying pathophysiological changes in the semi-acute post-injury period. Should a consistent pattern emerge that allows identification of patients at-risk for acute and/or persistent symptoms, such knowledge could guide development of therapeutic targets in mTBI and in understanding the most effective administration time window for these agents.