Time delay reconstruction for real systems is a widely explored area of nonlinear time series analysis. However, the majority of related work relates only to univariate time series, while multivariate time series data are common too. One such example is human gait kinematic data. The main goal of this article is to present a method of nonlinear analysis for kinematic time series. This nonlinear analysis is designed for detection of chaotic behavior. The presented approach also allows for the largest Lyapunov’s exponent estimation for kinematic time series. This factor helps in judging the stability of the examined system and its chaotic properties.