There is a pressing need in Australia and other countries to develop systems for monitoring secular trends in childhood obesity and related behavioural and environmental determinants. Energy from foods and beverages consumed at school is an accessible indicator of children's eating patterns and we have developed a school food checklist (SFC) to measure this. The SFC records the number of serves and source (home, canteen, vending machine) of 20 food and beverage categories. This study aims to assess the accuracy and to calibrate the SFC by comparing it to a weighed record (WR) and to evaluate inter-recorder reliability. Participants were 910 primary school children aged 5 to 12 years from a rural township in Victoria, Australia. WR were collected from a non-random sub-sample of 106 and a second sub-sample (n=46) had intake measured twice using the SFC to assess inter-recorder reliability. Mean energy values were 2992 kJ +/- 924 and 3008 kJ +/- 952 for the SFC and WR respectively and the correlation coefficient was strong (Pearson r = 0.77). The mean difference between the WR and SFC methods was 15 kJ (95% CI, -107 kJ to 138 kJ) and the limits of agreement (+2 standard deviations) were +/- 1270 kJ. The SFC overestimated the energy/serve of breads and fruit drinks and under-estimated energy/serve from fat spreads, biscuits/crackers, muesli/fruit bars and fruit. Inter-recorder reliability was good (kappa 0.51). The SFC was designed to measure energy from food and beverages in schools. It has good accuracy and reliability and the revised version should further improve accuracy of the instrument.