Currently environmental concerns are driving internal combustion engine manufacturers to seek greater fuel efficiency, more refinement and lower emissions. Cyclic variation is a known obstacle to achieving the greatest potential against these goals and therefore an understanding of how to reduce these is sought. It is widely accepted that cyclic variation in in-cylinder flow motions is a key contributor to overall cyclic variation and therefore the characterisation of factors affecting these is an important step in the process of achieving a better understanding and ultimately control of cyclic variation. This thesis reports the development of a novel optical engine research facility in which high speed digital particle image velocimetry (HSDPIV) has been applied to the study of flow field behaviour within a direct injection spark ignition (DISI) engine. This study investigates the spatial and temporal development of flow structures over and within many engine cycles. Flow field PIV measurements have been captured with a high spatial resolution and temporal frequencies up to 5 kHz from a number of measurement locations at a large range of crank angles. The major contributions from this work have included the use of the novel measurement technique to investigate spatial and temporal flow field development in the intake runner, valve jet, in-cylinder tumble and swirl planes and the pent roof. The gathered data have been used to investigate cycle by cycle variations in both high and low frequency flow structures. Major findings of this work have included the observation of highly varying flow fields throughout the engine cycle. Frequency analysis of these flows has allowed the low frequency bulk motions and higher frequency turbulent components to be studied. The low frequency flow field components are shown to create varying flow field interactions within the cylinder that also affect the manner in which the flow develops over the course of the cycle. The intensity of the turbulence fluctuations, u , has been calculated based upon the high frequency components within the flow and variations within this are shown to correlate with pressure related combustion parameters.