The Virtual Supermarket Task (VST) and Sea Hero Quest detect high-genetic-risk Alzheimer‘s disease (AD). We aimed to determine their test-retest reliability in a preclinical AD population. Over two time points, separated by an 18-month period, 59 cognitively healthy individuals underwent a neuropsychological and spatial navigation assessment. At baseline, participants were classified as low-genetic-risk of AD or high-genetic-risk of AD. We calculated two-way mixed effects intraclass correlation coefficients (ICC) for task parameters and used repeated measures ANOVAS to determine whether genetic risk or sex contributed to test-retest variability. The egocentric parameter of the VST measure showed the highest test–retest reliability (ICC = .72), followed by the SHQ distance travelled parameter (ICC = .50). Post hoc longitudinal analysis showed that boundary-based navigation predicts worsening episodic memory concerns in high-risk (F = 5.01, P = 0.03), but in not low-risk, AD candidates. The VST and the Sea Hero Quest produced parameters with acceptable test-retest reliability. Further research in larger sample sizes is desirable.