The objective of the research was to establish a scientifically-based decision making procedure for determining the amount of cash to be held at a cash point at any time without compromising the customer service level or incurring undue cost. To reach the objective, the problem was divided into the following subproblems: <ul> <li> To determine the cost parameters describing the nature of the problem of cash provision in South Africa.</li> <li> To investigate the characteristics unique to South African retail banking.</li> <li> To determine the nature of the demand distribution for a cash point.</li> <li> To develop a forecasting method appropriate for retail banking, although it was clearly stated that the methods used were specific to the branch studied.</li> <li> To investigate the existing order policies used by retail banks, as well as alternative order policies, with the aim of improving the cash replenishment process.</li> </ul> As a result of the investigation a generic decision model was developed which may be used to improve the process at branch level for retail banks in South Africa. Some suggestions were also made regarding the implementation and maintenance of the model. To investigate the cash replenishment problem, the cooperation of one of the leading retail banks in South Africa was obtained. A typical branch was selected. The total withdrawal, deposit patterns and the withdrawal patterns at the automated teller machines (ATM's) for a three month period during 1998 were investigated. The cost parameters relevant to the cash replenishment process were quantified. The approach followed was based on the classical inventory theory where the total cost of carrying inventory comprised three cost categories, i.e. storage cost, supply cost and shortage cost. Since the banks do not quantify the shortage cost, assumptions regarding the scope of the shortage cost had to be made. The next step was to determine the cost of the existing order policy followed by the branch. This figure was used as a benchmark once alternate policies were investigated. The investigation resulted in alternate policies which significantly reduced the daily cost involved in carrying inventory as well as reduced the average amount of cash carried at the branch. It was also shown, that the branch should consider using an appropriate forecasting method, since once forecasting was combined with an appropriate order policy, it was possible to reduce the cost of carrying cash inventories even further. In conclusion, the research report suggested an implementation plan to be followed at branch level pointing out that certain changes to information systems were required. In addition, training needs were identified to enable the branch operations manager to successfully use the decision support model. A comparison was drawn between the existing approach followed at the branch (which is mainly experience-based and largely of a random nature) to the proposed method. It was shown that the daily cost of carrying cash inventory could be reduced by 13 per cent per day. This represented a daily bottom line cost reduction ofR358. At the time that the research was carried out, this retail bank had 75 similar branches. Should the saving at this representative branch be extrapolated, it shows a potential saving of R8 000 000 per year at this category of branch. It was further shown that the average cash inventory at this branch could be reduced by 52 per cent using the proposed method. The study was limited to an investigation at one particular branch of a leading South African retail bank. The figures used to describe cash movements at the branch were of an extremely sensitive nature and were fairly difficult to obtain due to the way in which transactions are reported. The accuracy of the data provided by the branch could not be verified, but had to be accepted at face value. Although a particular case was investigated, a concerted effort was made to point out how the methodology may be used in the generic situation. During the period under review, the branch relocated to a complex across the street from its previous location in a busy shopping mall. This had a direct impact on the ATM withdrawal patterns at the two ATM's located at the branch. In addition, soon after the research was carried out, a number of other branches of the same retail bank were consolidated into this one particular branch. This would impact on the validity of the branch specific factors determined as part of the research. The study proved the applicability of industrial engineering principles in a service environment, where the added value of having the optimum cash amount available when required would impact directly on the bottom line of the bank and thereby enhance share-holder value. In the changing environment confronting retail banks, enhanced share-holder value is of the utmost importance to increase competitiveness and long-term survival.