Cancer patients with solid peripheral tumours suffer from concentration and memory symptoms as well as anxiety and depression. While other mechanisms are likely involved, neuroinflammation seems to play an integral role. Numerous studies have measured cytokines in the blood of patients as a proxy for neuroinflammation with the assumption that the serum inflammatory profile directly reflects the inflammatory profile of the brain. This approach to identify a common inflammatory culprit by blood assessment has yielded mixed results with some circulating cytokines being associated with cognitive symptoms in some studies whereas other studies report no association. To explore the extent to which changes in the blood also reflect changes in the brain and to identify which cytokines play a casual role in cancer-related cognitive impairment and psychiatric illness, we compared serum and brain cytokine profiles in three different mouse models of breast cancer. We show that the cytokine response to a mammary tumour is both tissue and brain region specific. Surprisingly, there were more dynamic changes in brain cytokine levels compared to serum, and these profiles differed for each brain region. Although changes in blood cytokine levels may not entirely reflect the brain cytokine response, the results do not rule out the use of cytokines as potential biomarkers for neurological symptoms in patients. The findings suggest that the lack of association between blood cytokines and cognition does not exclude an association between brain cytokines and cognition, and highlights the need for better strategies to accurately measure brain inflammatory profiles in patients. Given the little overlap in the serum and brain cytokine profiles between and within tumour models, it is plausible that the brain and body’s response to a peripheral tumour may not be generalizable. The findings have implications for how to apply cytokine therapies and promotes the development of personalised medicines.