Background Type 1 diabetes mellitus (T1D) affects the entire cellular network of the organism. Some patients develop cognitive disturbances due to the disease, but several authors have suggested that the brain develops compensatory mechanisms to minimize or prevent neuropsychological decline. The present study aimed to assess the effective connectivity underlying visuospatial working memory performance in young adults diagnosed with T1D using neuroimaging techniques (fMRI). Methods Fifteen T1D right-handed, young adults with sustained metabolic clinical stability and a control group matched by age, sex, and educational level voluntarily participated. All participants performed 2 visuospatial working memory tasks using a block design within an MRI scanner. Regions of interest and their signal values were obtained. Effective connectivity by means of structural equations models was evaluated for each group and task through maximum likelihood estimation, and the model with the best fit was chosen in each case. Results Compared to the control group, the patient group showed a significant reduction in brain activity in the two estimated networks (one for each group and task). The models of effective connectivity showed greater brain connectivity in healthy individuals, as well as a more complex network. T1D patients showed a pattern of connectivity mainly involving the cerebellum and the red nucleus. In contrast, the control group showed a connectivity network predominantly involving brain areas that are typically activated while individuals are performing working memory tasks. Conclusion Our results suggest a specific effective connectivity between the cerebellum and the red nucleus in T1D patients during working memory tasks, probably reflecting a compensatory mechanism to fulfill task demands.