Background: Reductions in pneumonia hospitalisations following introduction of pneumococcal conjugate vaccines (PCVs) have been reported from high-incidence and low-incidence settings but long-term data comparing vaccinated with unvaccinated children are sparse. Methods: We did a retrospective, population-based, record-linkage cohort study in Australian children using administrative health data from the Western Australian Midwives' Notification System and New South Wales Perinatal Data Collection, and the birth and death registries in both states. PCV vaccination details, pneumonia-coded hospital admissions, and invasive pneumococcal disease notification records were individually linked for children born between 2001 and 2012. The primary outcome was defined as the first hospital admission for all-cause pneumonia. Cox models were used to calculate adjusted hazard ratios (HR) to estimate the effect of PCV doses on pneumonia-coded hospital admissions by Aboriginal status, birth period, remoteness, and pneumonia diagnostic category in children younger than 2 years. Person-time of follow-up time for each child started at birth and was censored at the earliest of first hospital admission for all-cause pneumonia, death, invalid PCV dose, when the child reached age 24 months, or the end date of the study period (Dec 31, 2013) Findings: The study cohort comprised 1 365 893 children liveborn between Jan 1, 2001, and Dec 31, 2012, of whom 66 484 (4·9%) were identified as Aboriginal. The overall rate for all-cause pneumonia hospital admissions for children younger than 2 years over the entire study period was 17·6/1000 child-years in Aboriginal children and 5·5/1000 child-years in non-Aboriginal children. Compared with children born between 2001 and 2004 (ie, the pre-universal PCV period), the incidence of pneumonia-coded hospital admissions decreased in both vaccinated (6·5 vs 5·7 per 1000 child-years [12% reduction, 95% CI 3–21; p=0·01]) and unvaccinated non-Aboriginal children (6·8 vs 3·7 [45% reduction; 41–49]) born 2005–12 (the universal PCV period); among Aboriginal children, declines were significant only among those vaccinated (27·4 vs 14·1 [49% reduction, 40–55]). Among Aboriginal children born 2005–12, the risk of pneumonia-coded hospital admission after three doses of PCV was lower than those unvaccinated (adjusted HR 0·83, 95% CI 0·65–0·99) but, among non-Aboriginal children, the risk was similar (adjusted HR 1·09, 0·98–1·22). Overall, remote-born Aboriginal children had the highest incidence of hospital admission for pneumonia and among children born 2005–12, the adjusted risk was 37% lower (adjusted HR 0·63, 95% CI 0·42–0·96) among those fully vaccinated than those unvaccinated. Interpretation: Reductions in pneumonia-coded hospital admissions in unvaccinated children predominated in non-Aboriginal children with low incidence of pneumonia but were not significant in Aboriginal children with high incidence. These findings have potential implications for measuring PCV effect using a non-specific endpoint such as all-cause pneumonia in high-incidence populations. Funding: Commonwealth Government Collaborative Research Infrastructure Strategy and Education Investment Fund Super Science Initiative and the Australian National Health and Medical Research Council.