Chaotic behaviour of nonlinear systems has been one of the most fruitful additions to physics and applied mathematics in the latter part of the twentieth century. It has evolved into a mature discipline, which keeps researchers active in many areas of science. This special issue of the European Physical Journal is dedicated to reporting the latest advances in that area. The papers collected here span the full spectrum of the field, and show that it continues to be an exciting and active area of research. We have grouped all papers into four sections: a theoretical section, a section about synchronisation, a section on fluid dynamics, and a section on applications. We will now briefly address each paper and outline the most important results.Despite the fact that Dynamical Systems is a mature area of science, its fundaments and theoretical basis are still a source of many fascinating problems. One of the most exciting areas of dynamical systems theory is high-dimensional Hamiltonian dynamics. An important problem in the numerical study of Hamiltonian systems of high dimension is to efficiently compute the slow diffusion of trajectories near quasiperiodic tori. The first paper in this collection addresses this issue, and proposes an efficient method to calculate the GALI k coefficients, based on the singular value decomposition algorithm; the authors use their method to understand diffusion and other issues in the Fermi-Pasta-Ulam system [1].Another crucial area within the theory of dynamical systems deals with the development of time-series analysis techniques. Since Takens' theorem, embedding and time-series analysis have been the corner stones of many applications of dynamical systems theory. These topics also have a rich theoretical background, which is a research topic of its own. Vaidya and Majumder [2] investigate the cause of spurious instabilities often found in model equations reconstructed from time series data; they find that this is related to a topological foliation caused by the embedding. Henschel et al.[3] develop a multi-variate analysis method to adapt time-series analysis techniques for point processes -that is, data consisting of only a set of time points where a certain event happened, such as the firing of a neuron; this is crucial for the analysis of biomedical data. Kwasniok [4] tackles yet another aspect of time-series analysis: predictability. By fitting a hidden Markov chain model to time series data, he is able to predict the probabilities of visiting specific regions in phase space; the method is validated with the forced Lorenz system.Another major area in the theory of dynamical systems is bifurcation analysis. Non-smooth dynamical systems are well known for the rich set of bifurcations they display. Taralova [5] investigates a new bifurcation found in a particular class of piece-wise linear maps; this bifurcation leads to the creation of a chaotic orbit. A different kind of non-smoothness is studied by Pokorny and Klic in their paper [6], wherein they investigate the dynamics of a ...