Background: The semi-quantitative scale of structural brain Magnetic Resonance Imaging (sqMRI) is a valid and reliable measure of brain lesion extent in children with cerebral palsy (CP) >3-years. This system scores lesion burden for each major brain region. The sum of the scores gives a global score ranging from 0 to 48. Purpose: To investigate how sqMRI scores changed from infancy to school-age, and whether these were associated with lesion load, age at first assessment, and gross motor function and its changes. Materials and methods: Twenty-eight children with CP underwent MRI and motor (Gross Motor Function Measure-66; GMFM-66) assessments when <40-months and again when 8-12-years. We investigated whether (i) toddler/preschool-age sqMRI scores (Time 1) reflected school-age sqMRI scores (Time 2); (ii) temporal changes in sqMRI scores (Time 1–Time 2 difference) were related to the child's age at Time 1 and lesion extent; (iii) early or later sqMRI scores were associated with motor functioning; (iv) sqMRI scores' longitudinal changes were associated with motor changes. Results: Except for the corticosubcortical (grey-matter only) layers, sqMRI scores were significantly higher (‘higher lesion load’) at Time 1 than at Time 2. Age at Time 1 was not associated with temporal changes in global sqMRI scores. Higher lesion load at Time 2, but not at Time 1, was associated with smaller temporal changes in the global sqMRI score. The sqMRI scores were associated with concurrent, but not future or past motor GMFM-66 scores. Longitudinal changes in sqMRI scores were not associated with longitudinal changes in motor GMFM-66 scores. Conclusion: sqMRI scores of brain lesion extent at school-age are lower and a better indication of later-life motor functioning than very early life sqMRI scores. It may be best to interpret MRI white matter lesions with caution in very early life due to possible changes in lesion appearance and the unpredictable role of functional plasticity.