We aimed to evaluate the efficacy and safety of oxycodone/acetaminophen (O/A) and codeine/acetaminophen (C/A) vs. conventional therapy (CT) without opioids in older women suffering from osteoarthritis (OA)-related pain, sub-optimally responsive to prior conventional treatments. We performed a 6 week, randomized, single blind, controlled study in three nursing homes. We enrolled 154 women with painful OA. They were assigned to treatment with O/A (n=52) and C/A (n=52) vs. CT (n=50). We evaluated at baseline and at week 6: average pain in the last week (mean pain, MeP), pain at rest (RP), pain in movement (MP) (numeric rating scale, NRS); depressive symptoms (Beck Depression Inventory-II, BDI-II); functional status (activities of daily living, ADL) and cognitive status (mini mental state evaluation, MMSE). We considered the adverse events (AEs) in the study period. At week 6, MeP, RP and MP were significantly reduced in all three groups (p<0.001); compared to CT, O/A and C/A were associated with greater reductions in MeP (p<0.001 and p=0.004, respectively), in RP (p=0.028 and p=0.032, respectively) in MP (p<0.001 and p=0.002, respectively) and with significant improvement in BDI-II score (p=0.05 and p=0.04, respectively) and ADL value (p=0.04 and p=0.05, respectively). AE rates did not differ between groups.