Background: Increased oxycodone use has been associated with adverse drug events, non-medical use and overdose deaths. Objectives: To explore patterns of non-opioid, weak opioid and strong opioid use prior to initiation of oxycodone for non-cancer pain in a predominantly older Australian population. Methods: A retrospective study was conducted using the Australian Government Department of Veterans Affairs administrative claims database. Analgesic use 12 months prior to incident dispensing of oxycodone was determined for people in the community and in residential aged-care facilities (RACFs). Log-binomial regression was used to compute adjusted rate ratios (RRs) and 95 confidence intervals (95 CIs) for the use of other analgesics prior to initiating oxycodone. Results: Of 10,791 people who initiated oxycodone in 2010, 26 in community settings and 13 in RACFs were not dispensed other analgesics in the 12 months prior to initiating oxycodone. Thirty-four percent and 20 of those in community settings and RACFs, respectively, were not dispensed other analgesics in the previous 4 months. Co-morbidity had little impact on prior analgesic use. Each additional co-morbid condition was associated with a 1.4 increased likelihood (RR 1.014, 95 CI 1.012-1.016; p <0.0001) and a 1.2 increased likelihood (RR 1.012, 95 CI 1.009-1.015; p <0.0001) of being dispensed another analgesic prior to initiating oxycodone in community and RACF settings, respectively. Conclusions: Oxycodone is frequently initiated for non-cancer pain without first trialing other analgesics. This highlights the need for prescribing practices to be reviewed in light of increasing concerns about adverse drugs events and death due to oxycodone, particularly in older people.