Skin-markers based multi-segment models are growing in popularity to assess foot joint kinematics in different motor tasks. However, scarce is the current knowledge of the effect of high-energy motor tasks, such as running, on the repeatability of these measurements. This study aimed at assessing and comparing the inter-trial, inter-session, and inter-examiner repeatability of skin-markers based foot kinematic measures in walking and running in healthy adults. The repeatability of 24 kinematic measures from an established multi-segment foot model were assessed in two volunteers during multiple barefoot walking and running trials by four examiners in three sessions. Statistical Parametric Mapping (1D-SPM) analysis was performed to assess the degree of shape-similarity between patterns of kinematic measurements. The average inter-trial variability across measurements (deg) was 1.0 ± 0.3 and 0.8 ± 0.3, the inter-session was 3.9 ± 1.4 and 4.4 ± 1.5, and the inter-examiner was 5.4 ± 2.3 and 5.7 ± 2.2, respectively in walking and running. Inter-session variability was generally similar between the two motor tasks, but significantly larger in running for two kinematic measures (p < 0.01). Inter-examiner variability was generally larger than inter-trial and inter-session variability. While no significant differences in frame-by-frame offset variability was detected in foot kinematics between walking and running, 1D-SPM revealed that the shape of kinematic measurements was significantly affected by the motor task, with running being less repeatable than walking. Although confirmation on a larger population and with different kinematic protocols should be sought, attention should be paid in the interpretation of skin-markers based kinematics in running across sessions or involving multiple examiners.