Purpose: To examine the reproducibility of a triaxial seismic accelerometer under controlled conditions and real-life conditions. Methods: Instrumental reproducibility was examined using a shaker device. The accelerometers (DynaPort MiniMod; McRoberts B.V., The Hague, The Netherlands) were shaken at four frequencies (0.8, 2.1, 3.6, and 4.6 Hz) in x- and y-directions. The magnitude of acceleration ranged from 0 to 1.277g. Additionally, reproducibility under real-life conditions was examined in 55 adolescents (12-17 yr), with the accelerometer attached to the lower back. Each subject walked four short walking trials on level ground at preferred speed. To make this setting meet real-life conditions, we detached and reattached the accelerometer between trials 2 and 3. Detachment of accelerometer between trials 2 and 3 was done by either the same researcher or different researchers (four in total). Intra- and interobserver reproducibility were calculated. Results: Intra- and interinstrumental intraclass correlation coefficients (ICC) were 0.99 for both x- and y-directions. The intrainstrumental coefficients of variance (CoV) were lower than 1.13%. The interinstrumental CoV were lower than 1.37%. Intraobserver ICC was 0.97, and interobserver ICC was 0.88. Conclusion: The reproducibility of the accelerometer is high under the controlled conditions of a shaker device and in walking at preferred speed. © 2009 The American College of Sports Medicine.