This study seeks to make a contribution toward gaining a greater understanding of investment dynamics. The study has two main objectives: (1) to carry out a theoretical analysis of the dynamic behaviour of investment; and (2) to combine two strands of investment theory, dealing with financial constraints and financial development, and to apply the combined model to industry-level data. The study recognises that the neoclassical model represents a major advance from previous theories of investment, as it provides an explanation of how the desired capital stock is determined within a rigorous net-worth maximisation framework. By incorporating a convex adjustment cost function into that framework, the Q model has even greater theoretical appeal, in that it permits an explicit analysis of investment dynamics. However, the Q model generally does not perform well in empirical studies, due partly to a number of problems related to variable Q itself. To achieve the first objective, the dynamic properties of investment are analysed within the framework of a neoclassical model that, like the Q model, incorporates a convex adjustment cost function. However, rather than proceeding through first-order conditions involving the capital stock and its shadow price variable as in the Q model and many of its variants, this study employs the Euler equation approach to focus directly on the relationship between capital stock and investment, which is found to be characterised by saddle point equilibrium. This means that, following a shock (such as an unexpected cut in interest rates) the system must jump immediately to an appropriate point on the new saddle path associated with the new equilibrium. The study shows that under reasonable assumptions concerning the adjustment cost and production functions, these initial jumps typically involve investment overshooting its steady-state value in the short term. Investment overshooting also implies overshooting by stock prices in financial markets.