The diagnostic value of ultrasonography (US) for frozen shoulder (FS) is not well established. This study aimed to assess the diagnostic value of US measurement of inferior joint capsule (IJC) thickness and evaluate changes in the thickness of the IJC by US depending on arm position. A total of 71 patients with clinically diagnosed unilateral FS who underwent bilateral US measurement of the IJC were enrolled in this study. The US measurement of the IJC was performed with a linear transducer positioned around the anterior axillary line with the shoulder 40° abducted and with neutral rotation of the glenohumeral joint (neutral position). We also measured the IJC thickness in the externally rotated and internally rotated positions with the shoulder 40° abducted. In the neutral position, as well as in the internally rotated and externally rotated positions, the thickness of the IJC on US was significantly higher in the affected shoulder than that in the unaffected shoulder (all p < 0.001). On both the affected and unaffected sides, the US thickness of the IJC in the neutral position was significantly higher than that in the externally rotated position (p < 0.001), but lower than that in the internally rotated position (p < 0.001). Regarding IJC thickness in the neutral position, a 3.2-mm cutoff value yielded the highest diagnostic accuracy for FS, with a sensitivity and specificity of 73.2% and 77.5%, respectively. The area under the curve for IJC thickness was 0.824 (95% confidence interval, 0.76–0.89). US measurement of the IJC in the neutral position yielded good diagnostic accuracy for FS. Because IJC thickness is affected by arm rotation, it is important to measure the IJC thickness in a standardized posture to ensure diagnostic value.