This study is based on two observations about farmers' decisions to plant trees. The first is that it takes more labor to cut or uproot trees than to plant them. The second, is that rural labor markets are riddled with imperfections that make it difficult to hire labor for farm management so households tend to rely on family labor. The combination of these two factors may lead to a situation of economic irreversibility. This eventually locks land to trees. Farm households lose their ability to cope with risk ex-post because they have less flexible farming systems. The premise of this study is that households take these factors into account when choosing the rate, timing, and scale by which they plant trees. To verify this premise, a general model of natural resource accumulation under uncertainty and economic irreversibility was constructed. This model is a discrete time version of the common Dixit and Pindyck model. Stylized assumptions were used to conduct a numerical analysis. The objective of the numerical simulation is to derive general theoretical results and propositions about how economic irreversibility affects tree-planting decisions. Results of the analysis indicate that labor market imperfections lead to a lower volatility in the wage process. As seen from the stylized numerical simulation, this leads to earlier planting. Therefore, contrary to the findings of the standard model of exogenous irreversibility, the possibility of delayed timing is not true for economic irreversibility. The scale effects of irreversibility are also clearly seen from the analysis. Labor deficit or more labor-constrained households tend to plant early but will plant fewer trees per hectare. Numerical simulations showed that labor deficit households would plant faster compared to labor deficit households under economic irreversibility. It was also found that demographic variables (e.g. family size) affected the wage process and, therefore, the results can be attributed partly to these variables. Finally, the policy implication of this study is discussed. To be effective, agroforestry programs in the uplands must recognize heterogeneity of upland populations. As a consequence, the project rather than program approach is more appropriate when introducing upland interventions. These interventions must strongly consider the demographic characteristics and the position of the farming households in the labor market.