Objectives: The primary outcome was to study the utility of lumbar puncture (LP) in neonatal sepsis, and the secondary outcome was to study the clinical profile of the neonates with sepsis in a comparison of practices between two units. Methods: This is a retrospective comparative study conducted at two tertiary neonatal intensive care units (NICUs) during the period from January 2017 to January 2020. All neonates with C-reactive protein (CRP)-positive sepsis admitted to the NICU were included. Those with suspected sepsis but CRP-negative were excluded. The eligible medical records were reviewed, and data were recorded into a structured pro forma at both centers. The data from both units were analyzed. Results: Data from a total of 252 neonates were analyzed. Unit 1 (U1) had 29% blood culture-proven sepsis, of which 69% were late-onset, whereas at unit 2 (U2) it was 17.1%, of which 85% were late-onset and the remaining were early-onset. The incidence of meningitis was 3% in U1 and 1.9% in U2. Of these, 83% were associated with late-onset sepsis and only 17% with early-onset sepsis. Most of these neonates had the same organism in blood and cerebrospinal fluid (CSF). In those with meningitis, the highest CRP was <100 mg/L in 67% and >200 mg/L in 43% in U1, while all had a CRP <100 mg/L in U2. A receiver operating characteristic (ROC) curve plotted for CRP and culture-positive meningitis showed a CRP cutoff >80 mg/dL with a specificity of 84.9% and sensitivity of 66.7% for considering LP in sepsis. Conclusion: Lumbar puncture, being an invasive procedure, could be reserved only for those babies with culture-positive sepsis or with symptoms of meningitis or severe sepsis. This study suggests a CRP >80 mg/L as an indication for LP in neonatal sepsis.