We examined the influence of training on test administration accuracy and scoring reliability when physical activity instructors (PAIS) administered and scored the Fullerton Advanced Balance (FAB) scale. Test administration accuracy was evaluated with paired-samples t tests on seven non-clinician PAIs. Inter-rater reliability (IRR) on scoring was assessed using Spearman rank correlation coefficients among three non-clinician PAIs. Fisher's z transformation analysis was used to evaluate correlation differences in scoring reliability pre- to post-training. Test administration accuracy significantly improved post-training for six of ten test items. IRR was high both pre-training and post-training with correlations from .80 to 1.00 in 67% of the positive correlation pairings. No significant change in IRR pre- to post-training existed when comparing the total FAB test scores. Acceptable scoring reliability is achieved when PAIs utilize the published FAB scale test administration materials, and face-to-face training does not appear to improve scoring reliability when PAIs score the FAB as it is administered by an experienced test administrator. However, structured training does appear to improve test administration accuracy when the FAB scale is administered by PAIs.