This paper by Thomas Schreiber and Andreas Schmitz reviews and enhances the understanding of the limitations and caveats of surrogate data testing, a popular tool for assessing the necessity of applying nonlinear techniques to dynamical phenomena. The authors discuss various approaches to constrained randomization, including specific and general methods, and introduce new algorithms for handling unevenly sampled and multivariate data, as well as surrogate spike trains. They illustrate the main limitation, which lies in the interpretability of test results, through case studies and provide implementation details for these methods in the TISEAN software package. The paper also covers the detection of weak nonlinearity, higher-order statistics, phase space observables, and test design, emphasizing the importance of robust statistics over parametric methods. Additionally, it addresses issues such as periodicity artifacts and the general constrained randomization approach, which allows for more flexible null hypotheses and constraints.This paper by Thomas Schreiber and Andreas Schmitz reviews and enhances the understanding of the limitations and caveats of surrogate data testing, a popular tool for assessing the necessity of applying nonlinear techniques to dynamical phenomena. The authors discuss various approaches to constrained randomization, including specific and general methods, and introduce new algorithms for handling unevenly sampled and multivariate data, as well as surrogate spike trains. They illustrate the main limitation, which lies in the interpretability of test results, through case studies and provide implementation details for these methods in the TISEAN software package. The paper also covers the detection of weak nonlinearity, higher-order statistics, phase space observables, and test design, emphasizing the importance of robust statistics over parametric methods. Additionally, it addresses issues such as periodicity artifacts and the general constrained randomization approach, which allows for more flexible null hypotheses and constraints.