Background: Receiving vaccines at or close to their due date (vaccination timeliness) is a now key measure of program performance. However, studies comprehensively examining predictors of delayed infant vaccination are lacking. We aimed to identify predictors of short and longer-term delays in diphtheria-tetanus-pertussis (DTP) vaccination by dose number and ethnicity. Methods: Perinatal, notification, death and immunisation databases were linked for 1.3 million births in 2000–11 from two Australian states (Western Australia and New South Wales), with follow-up data until 2013. Ordinal logistic regression was used to estimate adjusted relative risks (RR) by degree of delay. Separate models were constructed for each vaccine dose and for Aboriginal and non-Aboriginal children. Results: Each dose-specific cohort included at least 49,000 Aboriginal and 1.1 million non-Aboriginal children. Delayed receipt was more common among Aboriginal than non-Aboriginal children (eg for the first dose of DTP [DTP1] 19.4 v 8.1%). Risk factors for delayed vaccination were strongest for DTP1, and delayed receipt of DTP1 was a key driver of subsequent delays; every week DTP1 was delayed was associated with a 1.6 to 2-fold increased risk of delayed DTP2 receipt. For DTP1, ≥3 previous pregnancies (the only factor more strongly associated with longer than shorter delays; RR ≥5 compared to no previous pregnancies), and children born to mothers <20 years of age (RR ≥2 compared to ≥35 years) were at highest risk of delay. Other independent predictors were prematurity, maternal smoking during pregnancy, and being born in Western Australia (if Aboriginal) or another country in the Oceania region. Conclusion: The sub-populations at risk for delayed vaccination we have identified are likely generalisable to other high-income settings. Measures to improve their dose 1 timeliness, particularly for children with older siblings, are likely to have significant flow-on benefits for timeliness of later doses.