Currently the main global emissions datasets (GCB, CDIAC, CEDS, etc.) do not include emissions from consumption of liquid fuels in Australia in the years before 1950 because of lacking data. As part of the Global Carbon Project's continued work to improve historical emissions estimates, this dataset presents physical quantities of petroleum commodities traded by Australia in the period 1903-1960. Given that the first commercial production of crude oil in Australia began after 1960 (source), trade data is sufficient to estimate consumption, on the assumption that stock changes were minimal. Petroleum reported as "ships' stores" are taken to be sales of international bunker fuels. While Australia did produce some liquid fuels in this period from oil shale, this is categorised as a solid fuel, and generally secondary products of solid fuels are also categorised as solid fuels, despite being liquid. The scale of production from oil shale is unknown, but assumed to be relatively small. Note that some commodity types present in the dataset may be assumed not to be oxidised, e.g. petroleum jelly, and therefore not contributing to CO2 emissions, and quantities of these commodities may not have been transcribed in all years. Further, for some years data for exports of domestically produced and foreign-produced commodities were transcribed separately, but for most years only the total exports were transcribed. The original PDF files from which data have been transcribed are also included. The data file indicates, for each record, the page number from which the record was obtained, counting the first page of the file as page 1 (as opposed to using printed page numbers). This allows verification, should that be desired. These reports are the annual "Oversea Trade" reports retrieved from the website of the Australian Bureau of Statistics (ABS) here. The 12-month period of reporting in most years is the Australian financial year, July through June, but in early years it was the calendar year. The 1913 edition was the final report for a calendar year, with the subsequent report being for the period July 1914 through June 1915. The latter report indicates that the six-month period between these two reports was published, but this was not located on the ABS website. The GCP's fossil CO2 dataset is documented here. Comparison of derived emissions with those published by CDIAC in the period 1950-1960 indicate good agreement.