Migraine is a heterogeneous disorder with variable symptoms and responsiveness to therapy. Current attempts to capture migraine variability through migraine subtyping are not informed by biology, ignore many migraine symptoms, and are not predictive of treatment responses. Taking advantage of neural network organization captured with resting-state functional connectivity (RSFC) and advanced statistical analysis, sophisticated symptom-brain mapping can now be performed. In aim one, I use a multivariate approach to relate clinical variability in migraine to RSFC, and find three dimensions of covariance between symptoms and the brain. Additionally, I show that the current subtyping of migraine does not adequately capture clinical heterogeneity. Instead, using the three identified dimensions of covariance, biotyping of migraine can be performed that does a better job of capturing migraine variability than the current field norm. In aim two I examine how RSFC can help to predict variability in migraine patient response to the mind-body therapy Mindfulness-Based Stress Reduction (MBSR), in the hopes of developing precision medicine for migraine. Finally, in aim three, I examine the mechanisms of MBSR by analyzing how MBSR changes functional connectivity to reduce the frequency of headaches. These findings suggest that novel approaches can better capture migraine variability, paving the way for the development of personalized treatment of migraine.