The cost of Parkinson's disease (PD) is associated with the progression of the disease. Levodopa is a potent treatment for this disease, but long-term treatment results in motor complications that are associated with both costs and considerable patient discomfort. The introduction of dopamine agonists early in the treatment of PD leads to a delay in these complications, but the treatment as such is associated with higher costs. We compared the cost-effectiveness of the dopamine agonist cabergoline (with the possibility of adding levodopa later) with that of levodopa alone. A Markov model with states of the disease without motor complications defined by Hoehn and Yahr stages, and a specific state for patients with motor complications was constructed. Transition probabilities between Hoehn and Yahr states were calculated from a previously published clinical trial using an ordered probit regression model. The risk of suffering from motor complications was estimated using a Weibull model. Mortality was assumed to be the same as in the general population. Costs were estimated based on a cross-sectional study performed in southern Sweden, and only direct costs were included. In the base case the model was run for 5 years (10 cycles) and extended to 10 years in a sensitivity analysis, and costs and effects were discounted at a rate of 3%. In the base case it was predicted that the cost in the levodopa arm was 7,453 euros (67,754 Swedish crowns) compared to 11,777 euros (107,065 crowns) in the cabergoline arm. However, patients treated with cabergoline gained 0.31 years without motor complications, giving an incremental cost-effectiveness ratio for cabergoline of 13,863 euros (126,023 crowns) per year of motor complications avoided (YMA). Running the model for 10 years resulted in a more favorable cost-effectiveness ratio: 6,122 euros/YMA (55,659 crowns). The model was robust to variations in discounting rates and to variations in the estimation of costs. Although no benchmark figure pertaining specifically to Parkinson's disease has been reported to suggest the economic value of a year without motor complications (gained as a result of treatment for Parkinson's disease), US$ 60,000 per QALY gained is a figure that is often used to assess cost-effectiveness in the literature. Using this criterion, cabergoline is cost-effective if the treatment saves 0.065 QALYs. Cabergoline is even more likely to be cost-effective when also considering indirect and informal costs. More research on the relationship between motor complications and quality of life is needed.