This paper describes the procedure followed to organise trade data obtained from the South African Revenue Service for inclusion in the PROVIDE Project Social Accounting Matrices. The data is aggregated in order to determine the custom's value of imports and exports, respectively, for each of the commodity groups identified in the Social Accounting Matrices. With regard to trading partners the trade data for South Africa is organized in such a way that it allows for the inclusion of multiple trading partners in the Social Accounting Matrices and that international trade at provincial level can be distinguished. Each data entry was mapped to a relevant commodity, province, and trading partner in three separate stages using three different mappings. General Algebraic Modelling Systems (GAMS) software was used to sort the trade data. Limitations of the data from the South African Revenue Service were addressed using additional information from Statistics South Africa, Global Insight and the South African Reserve Bank.