The purpose of this study was to evaluate the value of passive caudal gliding mobilization of the glenohumeral joint on the range of motion (ROM) of active and passive abduction; to evaluate the value of pain relief through visual analogue scale (VAS); to evaluate the correlation between improvement of shoulder abduction and intra-articular movement measured by fluoroscopy in frozen shoulder patients. The subjects consisted of twenty-one patients with clinically diagnosed frozen shoulder (11 males, 10 females) between 40 and 63 years of age (mean age : 52.7 years). The traction and caudal gliding mobilization based on the convex-concave rule in the resting position and at end range of abduction was peformed for 15 minutes per day and was repeated 10 times during a 2 week period. The ROM of abduction was measured by goniometer and pain was measured by VAS. The intra-articular movement was measured by fluoroscope, Neurostar Plus TOP (Siemens, Germany). ROM measurements of each patient was acquired at pre-treatment, immediate post-treatment and 2 week post-treatment. Statistical analysis was performed using SPSS 10.0 for Windows software and data was analyzed using the paired-test and the pearson correlation. The results of this study are as follows: 1. There was a significant decrease of VAS between pre-treatment data and 2 week post-treatment data (P<.05) but no significant difference between pre-treatment and immediate post-treatment data (P>.05). 2. There was a significant increase in ROM of active and passive abduction in the pre-treatment data, immediate post-treatment data, and in 2 week post-treatment data (P<.05). 3. With regard to results of the joint play test, there was a significant difference in the grade of traction between pre-treatment data and immediate post-treatment data and between pre-treatment data and 2 week post-treatment data (P<.05). There was no significant difference between immediate post-treatment data and 2 week post-treatment data (P>.05). 4. With regard to results of the joint play test, there was a significant difference in the grade of caudal gliding between pre-treatment data and immediate post-treatment data and between pre-treatment data and 2 week post-treatment data (P<.05). There was no significant difference between immediate post-treatment data and 2 week post-treatment data (P>.05), 5. With regard to the results of fluoroscopic findings, there was a significant change of the glenohumeral joint space between pre-treatment data and immediate post-treatment data and between immediate post-treatment data and 2 week post-treatment data (P<.05). There was no significant change of the glenohumeral joint space between immediate post-treatment data and 2 week post-treatment data (P>.05). 6. With regard to the results of fluoroscopic findings, there was a significant change of acromiohumeral joint space between the three data (pre-treatment data, immediate post-treatment data, 2 week post-treatment data) (P<.05). 7. Mobility grade by joint play test was significantly increased and was correlated to improved ROM of active and passive abduction (P<.05). In this study of frozen shoulder, passive caudal gliding techniques of the glenohumeral joint results in statistically significant changes in active and passive abduction as well as in VAS. There is also a significant correlation between joint play test and ROM of abduction.