Background: Diagnosis of dengue fever (DF) is challenging in the initial stage of illness. Early diagnosis and adequate management are important to reduce the complications associated with dengue. Objectives: This study aims to identify the clinical and laboratory features to predict DF from other febrile illnesses (OFI). Materials and methods: A observational analytical study was undertaken in an urban referral hospital in Mumbai, India. Eighty-seven children (up to 14 years of age) presenting with acute fever of >24 hours and <7 days without any evident or suspected focus on clinical examination were included. Clinical features and laboratory parameters at the time of presentation were used to build a predictive model for DF by multivariate logistic regression analysis. A rapid, qualitative immunochromatographic test for the detection of dengue non-structural protein-1 (NS1) antigen and IgG and IgM antibodies to dengue virus was done as a screening test for DF in all children. A serological test including the dengue IgM (MAC ELISA) test or its seroconversion after 2 weeks was considered as a diagnostic test for DF. Results: Dengue fever was diagnosed in 51.7% of children. Myalgia was an independent predictor of DF amongst the clinical features. A model including clinical and laboratory features demonstrated myalgia, leukopenia, and raised liver aspartate transaminase (AST) to be significant predictors of early DF. This model was internally validated, had high accuracy with an area under curve (AUC) of 0.93 (95% CI 0.88–0.98) with sensitivity 86.7% (95% CI 72.5–94.4), specificity 83.3% (95% CI 68.0–92.4), positive predictive value 84.78% (95% CI 70.51–93.16), and negative predictive value 85.4% (95% CI 70.13–93.9). Conclusion: We constructed a predictive model for the diagnosis of DF in an earlier stage of presentation. Validation of this model in a larger population and different regions should be attempted.