<p>Theories of embodied perception posit that the visuospatial perception of one’s environment is not only bound by features of the environmental itself, but also by the body capacity and affective state of the individual. Hill steepness and distance estimation tasks are widely used to assess the influence of physiological and psychological factors on visuospatial perception, but their test–retest reliability and measurement error are unknown. Such information is important to contextualise repeated measures study designs and understanding individual level differences. We aimed to evaluate the test–retest reliability and establish the smallest&#xa0;real&#xa0;difference (SRD) of three commonly used visuospatial perception tasks (ascending and descending hill steepness estimation, the distance-on-hill task) in healthy controls (<i>n</i> = 33) and people with painful knee osteoarthritis (<i>n</i> = 33). All participants completed the virtual reality visuospatial perception tasks two times, one week apart. Intraclass correlation coefficients (ICC), Bland–Altman plots, and SRD were used to evaluate the tasks. Our results revealed that both hill steepness estimation tasks are suitable for repeat administration in both populations given excellent reliability (uphill ICC = 0.80 to 0.85; downhill ICC = 0.89 to 0.90) and high sensitivity to change (uphill SRD = 17.7 to 18.9 degrees; downhill SRD = 12.1 to 14.7 degrees). The distance-on-hill task may have limited utility due to its poor reliability (ICC = 0.29 to 0.38) and low sensitivity to change (SRD = 6.20 to 8.5&#xa0;m). Our findings provide methodological support for the use of hill steepness tasks as a measure of visuospatial perception in embodied perception research.</p>