The aim of this study was to compare preoperative magnetic resonance imaging (MRI) findings, shoulder tests, and intraoperative observations in patients who underwent arthroscopic rotator cuff (RC) repair at a single center in Turkey. A total of 162 patients diagnosed with rotator cuff tears (RCTs) were included in this study. The physical examination involved the Jobe, Neer, and drop-arm tests. For patients undergoing shoulder arthroscopy, tear type and stage were evaluated intraoperatively and classified as partial-thickness or full-thickness tears. Preoperative assessments using MRI documented tear types and stages. Test results and imaging findings were compared to intraoperative observations to ensure diagnostic accuracy. A study of shoulder arthroscopy patients (129 right, 33 left; mean age 58.19 ± 9.20 years) found that 77.2% had full-thickness tears and 22.8% had partial tears on preoperative MRI. Significant differences were noted in the drop-arm test results (p = 0.003). MRI proved less reliable in localizing partial RCTs. Grade changes in full-thickness tears (Patte classification, p < 0.001) and partial tears (Ellman classification, p = 0.018) were statistically significant between preoperative and intraoperative evaluations. Certain specialized shoulder tests demonstrate strong diagnostic accuracy for specific conditions; however, their effectiveness may be limited when applied in isolation. MRI is a reliable tool for diagnosing RCTs; however, its sensitivity is reduced when detecting partial tears. This discrepancy suggests that both full-thickness and partial tears may exhibit more advanced pathology at the time of surgical intervention than initially indicated by preoperative imaging.