ObjectivesTo compare the effectiveness of oral versus intramuscular (IM) vitamin B12 (VB12) in patients aged ≥65 years with VB12 deficiency.DesignPragmatic, randomised, non-inferiority, multicentre trial in 22 primary healthcare centres in Madrid (Spain).Participants283 patients ≥65 years with VB12 deficiency were randomly assigned to oral (n=140) or IM (n=143) treatment arm.InterventionsThe IM arm received 1 mg VB12 on alternate days in weeks 1–2, 1 mg/week in weeks 3–8 and 1 mg/month in weeks 9–52. The oral arm received 1 mg/day in weeks 1–8 and 1 mg/week in weeks 9–52.Main outcomesSerum VB12 concentration normalisation (≥211 pg/mL) at 8, 26 and 52 weeks. Non-inferiority would be declared if the difference between arms is 10% or less. Secondary outcomes included symptoms, adverse events, adherence to treatment, quality of life, patient preferences and satisfaction.ResultsThe follow-up period (52 weeks) was completed by 229 patients (80.9%). At week 8, the percentage of patients in each arm who achieved normal B12 levels was well above 90%; the differences in this percentage between the oral and IM arm were −0.7% (133 out of 135 vs 129 out of 130; 95% CI: −3.2 to 1.8; p>0.999) by per-protocol (PPT) analysis and 4.8% (133 out of 140 vs 129 out of 143; 95% CI: −1.3 to 10.9; p=0.124) by intention-to-treat (ITT) analysis. At week 52, the percentage of patients who achieved normal B12 levels was 73.6% in the oral arm and 80.4% in the IM arm; these differences were −6.3% (103 out of 112 vs 115 out of 117; 95% CI: −11.9 to −0.1; p=0.025) and −6.8% (103 out of 140 vs 115 out of 143; 95% CI: −16.6 to 2.9; p=0.171), respectively. Factors affecting the success rate at week 52 were age, OR=0.95 (95% CI: 0.91 to 0.99) and having reached VB12 levels ≥281 pg/mL at week 8, OR=8.1 (95% CI: 2.4 to 27.3). Under a Bayesian framework, non-inferiority probabilities (Δ>−10%) at week 52 were 0.036 (PPT) and 0.060 (ITT). Quality of life and adverse effects were comparable across groups. 83.4% of patients preferred the oral route.ConclusionsOral administration was no less effective than IM administration at 8 weeks. Although differences were found between administration routes at week 52, the probability that the differences were below the non-inferiority threshold was very low.Trial registration numbersNCT 01476007; EUDRACT (2010-024129-20).