Capital flight is a serious concern for developing countries for its prevalence and its deleterious impacts on economic growth and welfare, macroeconomic stability, income distribution and other social costs, and linkage to other illegal activities. A good understanding of its determinants is essential for policymakers to formulate a sound response to the problems it imposes. This thesis first constructs a dataset of unrecorded capital flows, i.e. capital flight, for 35 developing countries over a time horizon of 21 years using the modified version of the residual measure. The inclusion of the trade misinvoicing component in the estimation of capital flight represents a major improvement over most of the previous studies. The estimation proves that capital flight is sizable. A corresponding dataset is constructed for net recorded capital flow, the analysis of which is meant to serve as a benchmark for the study of capital flight. Based on the literature review on some limitations of existing studies, the thesis employs dynamic models and fixed effects models in the empirical analysis of the determinants of international capital flow, with an aim to address the problems of autocorrelation and heterogeneity associated respectively with the inertia of capital flows and country specific effects. An innovation of the thesis is the utilization of the quasi-differencing technique in conjunction with dynamic and fixed effects models. The comparison of capital flight determinants and net recorded capital flow determinants is not known in previous studies and it proves informative in assisting the understanding of capital flight. Robustness tests are conducted by running regressions on sub-period samples as well as on two subsets of countries, namely the Latin American and Asian countries respectively. A very robust result is the significance of financial crises as a driver of capital flight. Capital inflows and financial deepening are also important factors. Asymmetric impacts of bank lending and trade on the recorded and unrecorded capital flows are also identified. Overall, the results seem to imply that a stable macroeconomic policy environment is the primary factor in the prevention of capital flight.