Over one third of Australians’ daily energy intake is from discretionary foods and drinks. While many health promotion efforts seek to limit discretionary food intake, the population health impact of reductions in the consumption of different types of discretionary foods (e.g., sugar-sweetened beverages (SSBs), confectionery, sweet biscuits) has not been quantified. This study estimated the potential reductions in body weight, obesity-related disease incidence, and healthcare cost savings associated with consumption of one less serving per week of different discretionary foods. Reductions in the different types of discretionary food were modelled individually to estimate the impact on energy consumption and population body weight by 5-year age and sex groups. It was assumed that one serving of discretionary food each week was replaced with either a serving of fruit or popcorn, and a serving (375 mL) of SSBs was replaced with coffee, tea, or milk. Proportional multi-state multiple-cohort Markov modelling estimated likely resultant health adjusted life years (HALYs) gained and healthcare costs saved over the lifetime of the 2010 Australian population. A reduction of one serving of SSBs (375 mL) had the greatest potential impact in terms of weight reduction, particularly in ages 19–24 years (mean 0.31 kg, 95% UI: 0.23 kg to 0.37 kg) and overall healthcare cost savings of AUD 793.4 million (95% UI: 589.1 M to 976.1 M). A decrease of one serving of sweet biscuits had the second largest potential impact on weight change overall, with healthcare cost savings of $640.7 M (95% CI: $402.6 M to $885.8 M) and the largest potential weight reduction amongst those aged 75 years and over (mean 0.21 kg, 95% UI: 0.14 kg to 0.27 kg). The results demonstrate that small reductions in discretionary food consumption are likely to have substantial health benefits at the population level. Moreover, the study highlights that policy responses to improve population diets may need to be tailored to target different types of foods for different population groups.