Aim Studies of published clinical trials involving children have shown frequent omissions in key aspects of design and conduct, but these problems may be artefactual and due to editorial processes and space limitations. To determine actual design and conduct, we analysed the completeness of key domains in trial protocols involving children submitted to Human Research Ethics Committees. Methods The ethics committees of all eight children's hospitals in Australia were invited to participate. De-identified trial protocols submitted for review in 2012 were evaluated using a checklist derived from Consolidated Standards of Reporting Trials, the Cochrane Risk of Bias Tool and Good Clinical Practice guidelines. Results Four ethics committees agreed to participate, and 69 protocols were analysed. The domains almost always reported were clustered around the background and trial plan (planned interventions for each group (99%), specific objectives (97%) and scientific background (96%)). Risk-of-bias domains such as random sequence generation and blinding of participants were often reported (75-90%). Domains least reported were clustered around the statistical analysis plan (66%), specified intention-to-treat analysis (54%), the justification for the proposed trial based upon a systematic review (48%) and age-specific outcomes (48%). Conclusions Protocols of trials involving children assessed by ethics committees generally include details on background and basic design, but many key domains in trial design and conduct are not covered. Despite widespread recognition of how problems in the design and conduct of trials may lead to unreliable results, investigators still appear to be omitting key elements in trial protocols.