We compared the short-term efficacy and safety of intra-articular (IA) botulinum toxin A (BoNT/A) to IA-placebo in patients with chronic, refractory shoulder joint pain. Forty-three shoulder joints in patients with moderate-to-severe shoulder arthritis pain were randomized to receive (1) 100 units IA-BoNT/A + lidocaine or (2) IA-saline + lidocaine. The following outcomes were compared using analysis of covariance: (1) primary: change in pain severity on a visual analog scale at 1 month (VAS, 0 cm to 10 cm); (2) secondary: Shoulder Pain and Disability Index (SPADI) disability subscale, quality of life on short-form (SF)-36 subscales, percent of patients who achieved at least a 30% decrease or a 2-point reduction in VAS pain (clinically meaningful pain relief), and safety. Both BoNT/A (n = 21) and placebo (n = 22) groups were comparable at baseline. At 1 month post-injection, the VAS pain reduction was significantly more in the BoNT/A group versus the placebo group (-2.4 vs -0.8; P-value = 0.014). When comparing BoNT/A with the placebo group at 1 month, it was observed that 5 SF-36 subscale scores improved significantly (P ≤ 0.035), and the SPADI disability improved more with a trend toward significance (51.5 ± 4.4 vs 64.9 ± 3.9; P = 0.083). In addition, clinically meaningful pain relief occurred in 61% versus 36% patients (P = 0.22). The total number of adverse events was similar, which included 50 events in the BoNT/A group versus 46 events in the placebo group. A single injection of BoNT/A produced statistically significant and clinically meaningful pain relief and improvement in quality of life in patients with chronic refractory moderate/severe shoulder arthritis pain at 1 month. These data provide evidence to support the efficacy of this novel neurotoxin therapy that needs to be confirmed in a multicenter, randomized trial.