Background: We conducted a prospective, single-centre, observational study to assess the serum chymase level, a mast cell derived protease, as a predictor of dengue severity. Methods: NS1-positive non-severe dengue patients of age > 14 years with duration of fever < 4days were included in the study. At the time of admission, the serum sample was taken for chymase estimation. Patients were followed up to four days after they became afebrile to find out the final diagnosis. Results: Total of 338 non-severe dengue patients were recruited (mean age - 29.15 years, male – 66%). On follow-up, 26 patients (7.8%) developed severe dengue. Only chymase level (adjusted odds ratio, aOR-1.787; 95%CI:1.309-2.440) and platelet count at admission (aOR-0.981; 95%CI:0.968-0.993) were able to predict the severity after adjustment for all variables. But, for prediction of severe dengue, the area under Receiver’s Operating Curve of chymase was 0.835 (95%CI:0.765-0.905), which was significantly higher than that of the platelet count at admission (0.760, 95%CI:0.650-0.870) (p < 0.001). Patients who developed severe dengue in due course of illness had significantly higher serum chymase level at admission as compared to the rest of the patients. Similar findings were noted across all age-groups. At an optimum cut-off value of 1.35 ng/mL, chymase had a positive likelihood ratio (LR) of 3.5 and a negative LR of 0.15, for predicting severe dengue. Conclusion: The results of this study are to be further validated by robust, multi-centric study designs. Controlled trials are warranted to assess whether a higher serum chymase level at admission could promote the allocation of healthcare resources to those who are most likely to develop severe dengue. Nonetheless, this study adds to the scarce literature of serum chymase as an objective predictor of dengue severity.