The relative cost-effectiveness of chloroquine (CQ) and sulfadoxine-pyrimethamine (SP) as first-line antimalarial therapy in southern Africa is of great interest to policymakers, clinicians and researchers in the subregion. A model was developed to access the cost-effectiveness of replacing CQ with SP as first-line treatment in Mpumalanga, South Africa, where malaria is seasonal and the population is non-immune. In-vivo drug resistance levels were used to derive a 'resistance variable' for each drug, which was used to compare the costs to the public healthcare provider associated with either therapeutic option. Costs including drugs, staff time, transport, maintenance, utilities, training and consumables were determined and subjected to Monte Carlo simulation and subsequent analysis to generate an average cost-effectiveness ratio (ACER) with confidence intervals for each drug. SP was found to be 4.8 (95% CI 3.3-6.7) times more cost-effective than CQ in Mpumalanga at 1997 resistance levels and costs, despite the far greater cost per treatment course of SP (US$ 4.02 as opposed to US$ 0.22 for CQ) in South Africa. At the price of SP in Kenya and Uganda (US$ 0.47-4.80 per treatment course), the ACER for SP does not change materially, increasing to between 5.1 and 5.6. Resistance emerged as the factor that most influenced the ACER of a specific drug. Indirect costs, compliance, changes in effectiveness and costs over time and costs of adverse events were not included in the model owing to paucity of data and logistical difficulties. Since most of these are likely to be similar in both drug models, the relative ACER is unlikely to be significantly altered by their inclusion.