Under the normal conditions of construction contracts, the client is obliged to pay the contractor in monthly instalments. The amount of each instalment is based on the value of construction work actually produced in the previous month and forecasts are needed in advance of the likely value of these payments. A database of previously completed contracts and payments made is available. A method for forecasting the value of these instalments is described. This method utilises three approaches, termed (1) analytic, (2) synthetic, (3) hybrid, in combination with six alternative models comprising (1) Hudson, (2) Kenley-Wilson, (3) Berny-Howes, (4) cumulative logistic, (5) cumulative normal, and (6) cumulative lognormal. The forecasts produced by each of these are then subject to a cross-validation analysis to determine the best approach/model combination for the available database and hence forecasts for future expenditure flows. An example is provided for an actual 27 construction project database.