Spatiotemporal gait analysis can provide quantitative information to assess treatment outcomes in stroke survivors. Therefore, clinicians need a portable, easy-to-use and low-cost tool that accurately and reliably measures spatiotemporal gait parameters. This paper examined the concurrent validity and reproducibility of the Gait Up gait analysis package for the spatiotemporal gait analysis in subacute stroke survivors. Twenty-five subacute stroke survivors participated in two walking tests. Spatiotemporal gait parameters were synchronously measured by 2 foot-worn inertial sensors (Physilog) and three-dimensional motion capturing (Vicon). Intraclass correlation coefficients, standard errors of measurement, smallest detectable changes, limits of agreement, and the Bland-Altman plots were calculated for the paretic and non-paretic side. After removing a consistent outlier (i.e., data of the paretic side of subject 36 who dragged his foot), agreement between both devices was good to excellent for paretic and non-paretic gait cycle time, cadence, stride length, stride velocity, and double support and moderate for paretic and non-paretic stance and swing. The Bland-Altman plots supported these findings. Test-retest reliability was good to excellent for most parameters, except paretic stance and swing. In conclusion, the Gait Up gait analysis package is a valid and reliable tool to measure paretic and non-paretic gait cycle time, cadence, stride length, and stride velocity in subacute patients with stroke, who do not exhibit severe dragging of the paretic foot. However, the algorithm should be improved for the analysis of paretic stance and swing phase.