BackgroundAlthough the clinical outcomes of manipulation under ultrasound-guided fifth and sixth cervical nerve root block for frozen shoulder have been reported, few studies have focused on the timing of manipulation. This study aimed to determine whether the timing of manipulation impacts the clinical outcomes. MethodsWe retrospectively reviewed the outcomes of 103 frozen shoulder patients (mean age 51.5 years) who underwent manipulation in one shoulder (n = 103 shoulders) between January 2012 and April 2019. Stiff shoulder was defined as limited range of motion in at least three directions, i.e., passive forward flexion of ≤100°, passive external rotation at the side of ≤10°, and internal rotation of ≤L5. The patients were categorized into two groups: those mobilized within 6 months after symptom onset (early group, 44 shoulders) and those mobilized >6 months after symptom onset (late group, 59 shoulders). The range of motion (forward flexion, external rotation, and internal rotation), Japanese Orthopaedic Association shoulder scores, Constant Shoulder Score, and University of California, Los Angeles scores before and 3, 6, and 12 months after manipulation were compared between groups. ResultsThe late group exhibited significant improvement in forward flexion, external rotation, internal rotation, Japanese Orthopaedic Association scores, Constant Shoulder Score, and University of California, Los Angeles scores at 3 months; forward flexion at 6 months; and forward flexion and University of California, Los Angeles scores at 12 months after manipulation compared to the early group. ConclusionsOur results indicate that timing has a significant influence on the outcome of manipulation for frozen shoulders. The optimal time for manipulation may be >6 months after symptom onset. These findings can be applied in counselling for frozen shoulder patients and for improved outcomes after manipulation.