In recent years fertility statistics in the ABS publication Births, Australia have suffered from a number of issues, including delayed receipt of registration data from the Registrars and computer processing errors. They prompt questions about the quality of these statistics for measuring the level of fertility and its year-to-year changes. The aim of this paper is to compare three sets of fertility statistics, (i) ABS birth registrations, (ii) ABS births by year of occurrence, and (iii) births in the Australian Institute of Health and Welfare National Perinatal Data Collection (NPDC), at the national and state/territory scales. Annual births by state/territory of usual residence from the three data sources were obtained for the period 2004-13 or 2004-14. Recast Estimated Resident Populations from the ABS were used to calculate Total Fertility Rates. For some states and territories there are non-trivial differences in fertility as recorded by the three data sources. In Queensland the trend in fertility according to ABS birth registrations is quite different from that based on the NPDC; in Tasmania and the Northern Territory birth registrations exceed NPDC births by a notable margin. There appears to be more uncertainty about the levels and annual changes in Australian fertility than many users of the data may realise. All three fertility datasets seem to possess some limitations. It is suggested that a new Australian fertility database be created which employs data linkage to incorporate both birth registrations and perinatal data.