Background: Prolonged fever contributes significantly as a health problem, and the etiology remains a diagnostic challenge in clinical practice, determined by factors including age-group, ethnicity, season, or geographic location. Using a structured approach for evaluation is useful. The aim of this study was to determine the etiology of prolonged fever in children using a modified four-stage protocol. Objectives: The primary objective was to determine the broad etiology of prolonged fever in children using a protocol-based approach. The secondary objective was to correlate common demographic and clinical parameters with the four etiological categories: infections, immune disorders, malignancy, and therapeutic trial or undetermined etiology. Materials and methods: A cross-sectional descriptive study design was used. Children aged 2 months to 18 years admitted with fever >7 days’ duration were recruited with consent. Baseline details and the modified four-stage protocol were used to identify the etiology of prolonged fever. Descriptive statistical analysis was done using SPSS 13. Results: 90 children, aged 3 months to 16 years (mean 6.3, SD 4.6), 60% male, with prolonged fever were included. Etiology was identified in 69 (76.7%) children, and infections were the commonest cause (51.1%), followed by inflammatory disorders (15.6%) and malignancy (10%). Rickettsial infection was the commonest (19.6%). Using the modified four-stage protocol, 33 children (36.7%) were identified in stage 1, 10 (11.1%) in stage 2, 21 (23.3%) in stage 3, and 26 (28.9%) remained undetermined or responded to therapeutic trial in stage 4. Conclusion: Infectious disease remains the most common etiology of prolonged fever in children, with anemia present across etiological groups. Relevant history, demographic information, symptoms, clinical examination, and baseline investigations are most important for diagnosis. Using a structured approach ensures judicious use of invasive and expensive investigations during evaluation.