Relationship-specific investment in land improvement (e.g. deep plowing and soil dressing) is necessary for upland farming. If farmers leasing farmland are in an environment where they are able to freely decide whether or not to invest in land improvement, no problem arises. However, if their intention to invest in land improvement is inhibited by some factors, the problem of inefficient farming due to underinvestment may arise. The problem of underinvestment is most likely to occur under off-the-record farming contracts (yami kosaku), which are a type of “incomplete contract” in the sense that the contract period is not predetermined. Because of incomplete contracts, farmers are unwilling to invest since they cannot predict if they will recoup their investment value while being open to eviction threats (i.e. holdup problem). Based on these points, this paper aims to analyze the empirical determinants on contract type (i.e. establishing right of use basis, or off-the-record contract basis) and the farmers'investment choice in land improvement, using original data from Atsumi-cho, Aichi Prefecture. The main findings are as follows. First, use-right contracts encourage farmers to invest in the leased land. It was indicated that the type of contract influences the incentive for land investment. Second, it was also found that the degree of trust (e.g. kinship and proximity) between the landowner and the farmer positively influences land investment. The third finding was that the degree of opportunity cost regarding the landowner's flexibility to engage in farming also influences contract choice. Landowners not likely to engage in agriculture were highly inclined to opt for use-right contracts. These findings call for the encouragement of use-right contracts by local governments, and a compensation scheme for “beneficial expenses” in order to provide farmers with incentives for land improvement.