Scoliosis patients were prospectively x-rayed in three positions with independent analysis. To determine if one positioning technique provides superior visualization of critical landmarks (C7, T2, T12, L5-S1) and to determine any position dependent variations in regional measures or sagittal balance. Different techniques for positioning patient's arms are used for 36" lateral radiograph with no data on relative effects. A total of 25 scoliosis patients were prospectively studied with 36" lateral radiographs in three positions varying arm location (straight out, partially flexed, and the "clavicle" position). Films were analyzed independently by three surgeons. Vertebral landmarks were scored for clarity; and lordosis, kyphosis, and global balance were analyzed. Statistical analysis was done with a General Estimating Equations model. The overall visualization score for the clavicle position was superior to either the 60 degrees or 90 degrees positions (clavicle vs. 60 degrees, P < 0.0001; clavicle vs. 90 degrees, P < 0.0003). Analysis of vertebral landmarks showed significantly better visualization of T2 with clavicle versus 90 degrees (P < 0.047), better visualization of T12 with clavicle versus either 60 degrees (P < 0.006) or 90 degrees (P < 0.049), and better visualization of L5-S1 with clavicle versus 90 degrees (P < 0.02). Regional measures showed no differences, but sagittal balance was significantly more positive in the 60 degrees position than either clavicle (P < 0.04) or 90 degrees (P < 0.015). The clavicle position for obtaining lateral 36" radiographs produces significantly better overall visualization of critical vertebral landmarks. Regional measures do not differ between the three positions, but global balance is more positive with the 60 degrees position. Clinically, the clavicle position may result in more accurate radiographic measures and may minimize repeated radiograph exposures.