Purpose: There has been a significant increase in the number of costly investigations of the shoulder joint over the past decade. The purposes of this study were to (1) describe the diagnostic imaging investigations ordered for injured workers seen at an Early Shoulder Physician Assessment (ESPA) program, (2) evaluate the impact of these investigations on final diagnosis and management, and (3) examine how efficient the program was by determining the appropriateness of referrals and whether costly imaging was justified. Methods: This was a retrospective review of the electronic files of injured workers who had been referred to an early assessment program because they had not progressed in their recovery or return-to-work plan within 16 weeks of the injury or reoccurrence. Results: The data of 750 consecutive patients-337 women (45%) and 413 men (55%), mean age 49 (SD 11) years-were reviewed. A total of 183 patients (24%) had been referred for further investigation. Of these, 90 (49%) were considered candidates for surgery (group 1), 58 (32%) had a change in diagnosis or management (group 2), and 17 (9%) had no change in diagnosis or management (group 3); 18 (10%) patients were lost to follow-up. We noticed a pattern in the type of diagnosis and the groups: full-thickness rotator cuff (RC) tear was the predominant diagnosis (Fisher's exact test [FET]=0.001, p<0.0001) for group 1. No statistically significant differences were found among the groups in the prevalence of labral pathology (FET=0.010, p=0.078), impingement syndrome (FET=0.012, p=0.570), partial-thickness RC tear (FET=0.004, p=0.089), or biceps pathology (FET=0.070, p=0.149). Ultrasound investigations were more prevalent in group 2 (FET=0.004, p=0.047). No pattern was found for use of magnetic resonance imaging and group allocation. However, all magnetic resonance arthrogram investigations (FET=0.007, p=0.027) had been ordered for patients who required labral or instability-related surgery. Conclusions: Of the injured workers we studied, 24% had further investigation, and the type and severity of pathology had affected the type of investigation. For the 165 patients who were included in groups 1-3, the ESPA was 90% efficient, with only 10% of patients not having had a change in diagnosis or management.