Objective: To evaluate the processes by which pharmaceuticals are added to the formularies of Australian paediatric hospitals. Design: Descriptive study of the processes and outcomes of all submissions to Australian paediatric hospital drug and therapeutics committees from 1 July 2010 to 31 December 2011. Setting: All eight tertiary paediatric hospitals in Australia. Participants: Interviews with committee secretaries or delegates and document analysis. Main outcome measures: Total number of formulary applications, stratified by therapeutic class, approval rates for each hospital and quality of supporting information. Results: One hundred and twenty applications were considered during the study period, with most applications approved (range, 67%-100%). Neurological agents were the most common therapeutic class considered. A conflict of interest was declared for 10 applications (8%). Forty-five (38%) were independently reviewed by a statewide medicines advisory committee or hospital pharmacist. Several committees approved identical applications during the period of review and with different outcomes. For applications submitted for new drugs or new indications (95 applications), supporting data included randomised controlled trials (37/95), case series (36/95), product information (34/95) and narrative reviews (29/95). Few applications (14/95) included a systematic review or meta-analysis. No application included an evaluation of the risk of bias of supporting studies. Conclusions: There is limited high-quality evidence informing paediatric hospital-based drug approvals. Approval processes vary considerably among institutions with substantial duplication of effort and variable outcomes. Resources and training appear insufficient given the technical complexity of submissions. A national, standardised approach to hospital-based drug evaluation could reduce overlap and improve decision making.