To determine the diagnostic performance (ie, sensitivity, specificity, interrater reliability) of the thumb metacarpal adduction and extension tests against traditional examination maneuvers for trapeziometacarpal (TMC) arthritis. This cross-sectional study recruited 129 patients from 2 outpatient offices at a tertiary institution. All patients had radiographic wrist examinations and completed a standardized physical examination consisting of the thumb adduction and extension tests as well as standard examination maneuvers for radial wrist and thumb pain. The physical examinations were performed by 1 of 2 attending physicians and an independent examiner. Patients were recruited for 3 diagnostic groups: TMC arthritis, radial wrist or hand pain, and nonradial wrist pain controls. Statistical analysis calculated the sensitivity, specificity, and interrater reliability of each physical examination maneuver for detecting TMC arthritis. The thumb adduction maneuver was found to have a sensitivity of 0.94 (confidence interval [CI], 0.82-0.98) and a specificity of 0.93 (CI, 0.86-0.97). The thumb extension maneuver had a sensitivity of 0.94 (CI, 0.82-0.98) and a specificity of 0.95 (CI, 0.87-0.98). The interrater reliability was excellent for both the adduction (κ = 0.79) and the extension tests (κ = 0.84). The grind test had a sensitivity of 0.44 (CI, 0.30-0.59), a specificity of 0.92 (CI, 0.84-0.97), and poor interrater reliability (0.31). Point tenderness at the TMC joint had a sensitivity of 0.94 (CI, 0.82-0.98), a specificity of 0.81 (CI, 0.71-0.88) and fair interrater reliability (κ = 0.63). The adduction and extension tests each proved to be more sensitive than the grind test for the detection of TMC arthritis. Further, these provocative tests were more specific for basal joint arthrosis than was the elicitation of point tenderness at the joint. The metacarpal adduction and extension maneuvers demonstrated excellent utility as screening tests for the identification of TMC arthritis. Diagnostic II.