OBJECTIVE To optimise image quality and reduce radiation exposure for infant body CT imaging. METHODS An image quality CT phantom was created to model the infant body habitus. Image noise, spatial resolution, low contrast detectability and tube current modulation (TCM) were measured after adjusting CT protocol parameters. Reconstruction method (FBP, hybrid iterative and model-based iterative), image quality reference parameter, helical pitch and beam collimation were systematically investigated for their influence on image quality and radiation output. RESULTS Both spatial and low contrast resolution were significantly improved with model-based iterative reconstruction (p < 0.05). A change in the helical pitch from 0.969 to 1.375 resulted in a 23% reduction in total TCM, while a change in collimation from 20 to 40 mm resulted in a 46% TCM reduction. Image noise and radiation output were both unaffected by changes in collimation, while an increase in pitch enabled a dose length product reduction of ~6% at equivalent noise. An optimised protocol with ~30% dose reduction was identified using model-based iterative reconstruction. CONCLUSIONS CT technology continues to evolve and require protocol redesign. This work provides an example of how an infant-specific phantom is essential for leveraging this technology to maintain image quality while reducing radiation exposure. CONCLUSIONS • A size-specific phantom is critical in protocol development for infant CT. • New reconstruction technology enables ~30% dose reduction at equivalent image quality. • A consistent performance is observed for this scanner system across protocol changes. • A tradeoff exists between reducing exposure time and enabling tube current modulation.