AIM: It was recently reported that serum retinol-binding protein (RBP), also known as retinol-binding protein 4 (RBP4), was positively associated with systemic insulin resistance. We hypothesized that an imbalance between RBP and retinol might be the underlying cause for this association. METHODS: We studied the ratio between RBP and retinol in 233 humans divided into groups depending on normal glucose tolerance (NGT), impaired glucose tolerance (IGT), type 2 diabetes (T2DM) and presence or absence of obesity. RESULTS: Plasma RBP and retinol levels were lower in patients with T2DM than in individuals with NGT (p < 0.05 and p < 0.0001 respectively). In contrast, RBP-to-retinol ratio was higher in individuals with T2DM (p < 0.0001) and IGT (p < 0.05). Following multivariate adjustment, RBP and retinol correlated positively with low-density lipoprotein (LDL) and triglycerides (p < 0.0001, except retinol and LDL: p < 0.001). RBP-to-retinol ratio correlated positively with glucose 2 h after an oral glucose tolerance test (p < 0.0001) and with C-reactive protein (p < 0.001). Retinol, RBP and adipose tissue RBP messenger RNA (mRNA) levels shared an inverse relationship with plasma interleukin-6, and adipose tissue RBP mRNA levels correlated positively with plasma tumour necrosis factor-alpha (TNF-alpha) and skeletal muscle TNF-alpha mRNA levels. CONCLUSIONS: Our results suggest that the excess of RBP relative to retinol, assessed as the RBP-to-retinol ratio, is more indicative of T2DM than RBP itself. Hence, the previously reported insulin resistance in mice induced by overexpression or injection of RBP could be because of higher levels of RBP relative to retinol rather than higher total levels of RBP. Moreover, TNF-alpha may have a role in RBP-mediated adipose to muscle crosstalk.