Inflammatory processes as well as attenuation of brain-derived neurotrophic factor (BDNF) availability are involved in the pathophysiology of major depressive disorder (MDD). Although it is generally presumed that these two systems interact negatively in the brain, preclinical and human in vitro studies have shown synergistic rather than antagonistic interactions in the periphery. We therefore examined the association between serum levels of BDNF and plasma levels of interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-α) in patients with MDD (n=1070) and non-depressed controls (n=379) from the Netherlands Study of Depression and Anxiety. We used multiple regression analyses with serum BDNF as the dependent variable and we specifically tested the presence of BDNF-cytokine associations in DSM-IV-assigned melancholic MDD patients, identified by the Inventory of Depressive Symptomatology. After adjustment for sociodemographics, sampling variability, lifestyle indicators, somatic diseases and medication use, BDNF levels were predicted by the interaction between MDD diagnosis and IL-6 (p-interaction=.006). Stratified analyses showed that BDNF levels are indeed positively associated with IL-6 levels in MDD patients (β=.07, p=.02), but not in non-depressed controls (β=-.07, p=.23). When further stratified for melancholic and non-melancholic MDD (p-interaction=.005), IL-6 emerged as a robust positive predictor of BDNF only in the melancholic sample (β=.21, p=.01), wherein serum BDNF levels were accordingly enhanced. Post-hoc exploratory analyses verified an accentuated positive association of BDNF levels with leucocyte counts in melancholia. No significant associations emerged between BDNF and TNF-α. Overall, our cross-sectional approach may have disclosed an allostatic, BDNF-inducing component of peripheral immunity and/or an immunotrophic function of peripheral BDNF. Both scenarios may warrant further exploration, as they could inform new research concepts towards immune-based antidepressive treatment strategies.