Demand uncertainty is a significant issue faced by both the government and private firms in build-operate-transfer (BOT) toll road projects. To encourage private firms' participation, the government usually provides them with demand guarantee. This paper investigates and compares the optimal BOT contracts with the minimum demand guarantee (MDG) and the flexible demand guarantee (FDG), respectively, in environments of symmetric and asymmetric information. It is determined that under the MDG with asymmetric information, the private firm's optimal effort decreases with respect to the guarantee level, which indicates that the MDG provides the private firm with a disincentive to exert effort to increase traffic demand. Under the FDG with asymmetric information, the private firm's optimal effort decreases with respect to the guarantee level and threshold coefficient. Through a comparison of the environment with asymmetric information, the optimal toll price, optimal guarantee level and resulting social welfare are higher under the FDG for a small threshold coefficient. Therefore, it is concluded that the government may choose the FDG instead of the MDG for a small threshold coefficient, which means that the FDG is suitable for toll roads with relatively low demand. Conversely, the government prefers to choose the MDG instead of the FDG when the threshold coefficient is large enough, which means that the MDG is suitable for toll roads with relatively high demand. We further demonstrate that our model results still qualitatively hold when traffic congestion is considered.