Inflammatory illness is associated with depression. Preclinical work has shown that chemokines are linked with peripheral-central crosstalk and may be important in mediating depressive behaviours. We sought to establish what evidence exists that differences in blood or cerebrospinal fluid chemokine concentration discriminate between individuals with depression and those without. Following PRISMA guidelines, we systematically searched Embase, PsycINFO and Medline databases. We included participants with physical illness for subgroup analysis, and excluded participants with comorbid psychiatric diagnoses. Seventy-three studies met the inclusion criteria for the meta-analysis. Individuals with depression had higher levels of blood CXCL4 and CXCL7 and lower levels of blood CCL4. Sensitivity analysis of studies with only physically healthy participants identified higher blood levels of CCL2, CCL3, CCL11, CXCL7 and CXCL8 and lower blood levels of CCL4. All other chemokines examined did not reveal significant differences (blood CCL5, CCL7, CXCL9, CXCL10 and cerebrospinal fluid CXCL8 and CXCL10). Analysis of the clinical utility of the effect size of plasma CXCL8 in healthy individuals found a negative predictive value 93.5%, given the population prevalence of depression of 10%. Overall, our meta-analysis finds evidence linking abnormalities of blood chemokines with depression in humans. Furthermore, we have demonstrated the possibility of classifying individuals with depression based on their inflammatory biomarker profile. Future research should explore putative mechanisms underlying this association, attempt to replicate existing findings in larger populations and aim to develop new diagnostic and therapeutic strategies.