Reliable markers for identifying infections in cancer patients on admission are lacking. The utility of the balance between interleukin (IL)-10 and IL-12 was analysed in this respect. The infection group (n=56) had higher median serum levels of IL-10 (3.8 pg/ml; interquartile range (IQR) 1.7–11.4 pg/ml versus 1.8 pg/ml; IQR 0.6–4.6 pg/ml; P=0.005) and IL-10 to IL-12 ratio (0.4; IQR 0.06–4.23pg/ml versus 0.05; IQR 0.02–0.31pg/ml; P<0.001) than the non-infection group (n=36). IL-10 and the ratio had the following figures of sensitivity (79%; 95% confidence interval (CI) 66–88 versus 39%; 95% CI 27–53), specificity (40%; 95% CI 12–74 versus 90%; 95% CI 56–100) and positive predictive value (88%; 95% CI 76–96 versus 96%; 95% CI 78–100) for identifying infections (56 cases with infection and 10 with neoplastic fever), and the corresponding area under curve (AUC) values for IL-10 and the ratio in identifying infections in general were 0.58; 95% CI 0.39–0.78 versus 0.64; 95% CI 0.46–0.82 and in bacteraemia 0.71; 95% CI 0.50–0.92 versus 0.75; 95% CI 0.58–0.93, respectively. Thus, IL-10 can be used as a screening method for identifying infections in cancer patients and the ratio of IL-10 to IL-12 for confirming the diagnosis.