Introduction: The purpose of this study was to determine if demographic variables and measures of cognitive function, functional mobility, self-reported balance self-efficacy, and self-reported physical activity can predict gait speed variability during single-task walking (STGSCV), during cognitive-motor dual-tasking (DTGSCV), and dual-task effect on gait speed variability (DTEGSCV) in older adults. Methods: In 62 older adults, demographics were recorded and cognitive function (including the Montreal Cognitive Assessment, MoCA), functional mobility, balance self-efficacy (Activities-specific Balance Confidence Scale, ABC), and self-reported physical activity (Physical Activity Scale for the Elderly, PASE) were assessed. Three linear regression models were used to determine whether the functional mobility tests, PASE score, ABC score, and tests of cognitive function predicted gait speed variability outcomes (STGSCV, DTGSCV, DTEGSCV), with demographics included as covariates. Results: MoCA score (p = 0.003), gender (p = 0.040), and years of education (p = 0.010) significantly predicted DTGSCV (R2 = 0.297, p = 0.002). MoCA score (p = 0.008) and years of education (p < 0.001) also significantly predicted DTEGSCV. Despite a significant regression model (R2 = 0.316, p = 0.001), there were no significant individual predictors of STGSCV. Conclusion: Older adults with lower cognitive function scores, according to the MoCA, and higher levels of education demonstrate larger gait speed variability during dual-tasking and may be at increased risk of real-world mobility issues or falling.