The use of musculoskeletal ultrasonography (MSUS) in guiding subdeltoid injection has been shown to improve outcome up to 6 weeks in a few small studies. A recent meta-analysis identified the need for further studies with longer-term outcome and larger sample size. This randomized prospective study assessed whether clinic-based MSUS can significantly improve diagnostic accuracy in shoulder pain and whether MSUS-guided shoulder injection results in improved long-term outcomes. One hundred consecutive patients with 125 painful shoulders were recruited. Patients were randomized to receive either sonographic assessment with consequent palpation-guided injection (Group 1, n = 66) or sonographic assessment with a MSUS-guided injection of 40 mg of methylprednisolone acetate (Group 2, n = 59). A blinded rheumatologist (ADF) performed clinical assessments at baseline, 6 and 12 weeks including shoulder function tests (SFTs) (Hawkins-Kennedy test, supraspinatus tendon tenderness), physician global assessment (PGA) and patient visual analogue scores (VAS) for pain (0-10). Eighty patients with 90 symptomatic shoulders completed 12-week follow-up. Twenty patients, 11 (20 shoulders) from the palpation-guided group and 9 (15 shoulders) from the MSUS-guided group, were excluded at 6 weeks either due to requirement for repeat injection or due to surgical referral. Mean age for patients was 57.7 years, and 65 % patients were female; mean shoulder pain duration was 18 weeks (range 14-22 weeks). SFTs, patient VAS and PGA scores for pain improved significantly from baseline in both groups with significantly greater improvements in the MSUS-guided group (44 shoulders) compared to the palpation-guided group (46 shoulders) in all parameters at 6 (p < 0.01) and 12 weeks (p < 0.05). The use of MSUS in guiding subdeltoid injection has been shown to improve outcome up to 6 weeks in a few small studies. A recent meta-analysis identified the need for further studies with longer-term outcome and larger sample size.