To determine the effect of patient age on the accuracy of primary MRI signs of long head of biceps (LHB) tendon tearing and instability in the shoulder using arthroscopy as a reference standard. Subjects with MRI studies and subsequent arthroscopy documenting LHB tendon pathology were identified and organized into three age groups (18-40, 41-60, 61-87). Normal and tendinopathic tendons were labeled grade 0, partial tears grade 1 and full tears grade 2. Two radiologists blinded to arthroscopic data graded MRI studies independently. Prevalence of disease, MRI accuracy for outcomes of interest, and inter-reader agreement were calculated. Eighty-nine subjects fulfilled inclusion criteria with 36 grade 0, 36 grade 1 and 17 grade 2 tendons found at arthroscopy. MRI sensitivity, regardless of age, ranged between 67-86% for grade 0, 72-94% for grade 1 and 82-94% for grade 2 tendons. Specificity ranged between 83-96% for grade 0, 75-85% for grade 1 and 99-100% for grade 2 tendons. MRI accuracy for detection of each LHB category was calculated for each age group. MRI was found to be least sensitive for grade 0 and 1 LHB tendons in the middle-aged group with sensitivity between 55-85% for grade 0 and 53-88% for grade 1 tendons. Agreement between MRI readers was moderate with an unweighted kappa statistic of 62%. MRI accuracy was moderate to excellent and agreement between MRI readers was moderate. MRI appears to be less accurate in characterizing lower grades of LHB tendon disease in middle-aged subjects.