prediction include earnings, accrual terms such as changes inaccount receivable and payable and disaggregated cash compo-nents. Empirical studies suggest that these variables are usefuland informative in predicting cash flows. In this paper, a compar-ison is made between different sets of variables as predictors. Itis expected that the more predictors that are included, the bettera model will perform because more inclusion of variables oftenmeans richer exploitation of information. A second comparisonis made between models which include explanatory variableswith different lags. It is expected that models with more laggedexplanatory variables could provide more accurate predictionwhereas the reported results in this paper suggest otherwise. Thispaper proposes two types of cash flow modeling, i.e. movingaverage model and vector autoregressive (VAR) model for cashflow prediction. Moving average model also makes economicsense as it measures how an unexpected cash flow shock couldinfluence people making future prediction. The moving averagemodel is applied to one-period-ahead prediction and VAR modelis proposed for multi-period-ahead prediction. For this purposeVAR is more powerful and relies less on data availability thanlinear regression. These models are applied empirically on dataof South Africa firms.This paper is organized as follows: Section 1 provides theintroduction to this paper. Section 2 reviews the literature anddiscusses factors that influence the prediction of cash flows andthe prediction models utilized in the study. Section 3 describesthe data for South African firms. Section 4 reports the results ofthe empirical analysis and the conclusion of the study is providedin Section