The purpose of this study was to determine the 1-week test-retest reliability of the Sensory Organization Test (SOT), using computer-generated scores and loss of balance (LOB) episodes in noninstitutionalized older adults. The SOT was administered to each subject on two separate days 1 week apart in an out-patient clinic. A volunteer sample of 40 individuals who were at least 65 years of age participated in this study. The main outcome measures were computer-generated scores for the first trial and the average of the three trials in each of the six sensory conditions of the SOT; computer-generated composite score of the six conditions; LOB on the first trial and any of three trials in each condition. The intraclass correlation coefficients (ICC) for the SOT first trial data ranged from .15 in Condition 3 to 0.70 in Condition 5. The ICCs for the SOT average of three trials ranged from 0.26 in Condition 3 to 0.68 and 0.64 in Conditions 5 and 6. Percent agreement was 77% to 100% for LOB on the first trial, as well as LOB on any of three trials of Conditions 1 through 6. As the conditions became more difficult, an increasing number of subjects experienced LOB. Analysis revealed fair to good test-retest reliability for computer-generated scores and good reliability for LOB across some conditions of the SOT. A modification to the current scoring system is suggested which would improve the reliability of the computer-generated scores of the SOT.