In recent years, pediatric research on tuberculosis (TB) has focused on addressing new biomarkers with the potential to be used as immunological non-sputum-based methods for the diagnosis of TB in children. The aim of this study was to characterize a set of cytokines and a series of individual factors (ferritin, 25-hydroxyvitamin D [25(OH)D], parasite infections, and nutritional status) to assess different patterns for discriminating between active TB and latent TB infection (LTBI) in children. The levels of 13 cytokines in QuantiFERON-TB Gold In-Tube (QFT-GIT) supernatants were analyzed in 166 children: 74 with active TB, 37 with LTBI, and 55 uninfected controls. All cytokines were quantified using Luminex or ELISA. Ferritin and 25(OH)D were also evaluated using CLIA, and Toxocara canis Ig-G antibodies were detected with a commercial ELISA kit. The combination of IP-10, IFN-γ, ferritin, and 25(OH)D achieved the best diagnostic performance to discriminate between active TB and LTBI cases in children in relation to the area under receiver operating characteristic (ROC) curve 0.955 (confidence interval 95%: 0.91–1.00), achieving optimal sensitivity and specificity for the development of a new test (93.2 and 90.0%, respectively). Children with TB showed higher ferritin levels and an inverse correlation between 25(OH)D and IFN-γ levels. The model proposed includes a combination of biomarkers for discriminating between active TB and LTBI in children to improve the accuracy of TB diagnosis in children. This combination of biomarkers might have potential for identifying the onset of primary TB in children.