A prospective observational study was conducted on 500 children to evaluate the Integrated Management of Neonatal and Childhood Illnesses (IMNCI) algorithm for diagnosis and referral of children aged 0–59 mo. IMNCI algorithm performed well in diagnosis [except sepsis in 0–7 d (p < 0.01) and malaria in 2–59 mo (p < 0.01)] and referral of under five illnesses. The proportion of children with the mismatch between ‘IMNCI’ diagnosis and ‘Gold Standard’ diagnosis were highest (38.7 %) among children 0–7 d of age. The authors conclude that IMNCI algorithm in young children has good sensitivity for referring children with severe illnesses and is a good tool for diagnosis of most of childhood illnesses in under-five children.