Abstract: To investigate the distribution of tobacco leaf's moisture content in the whole space of bulk curing barns and to clarify the dynamic change of the moisture content during leaf curing process, six parameters were selected including the flue-curing time, ambient temperature, relative humidity, mass of the tobacco monitoring unit, the distance between the cross sections of the tobacco monitoring unit and the air outlet, and the height of the tobacco monitoring unit from the ground were selected as input characteristics, while the moisture content in the tobacco leaves was selected as an output parameter. A SSA-BP model was established to predict the moisture content distribution of tobacco leaves at different positions in the bulk curing barn. The results showed that: 1) The SSA-BP prediction model was more accurate in predicting the moisture content of tobacco leaves in the bulk curing barn. 2) The feature importance analysis based on the CART decision tree algorithm showed that the spatial location parameters (the distance between cross sections of the tobacco monitoring unit and the air outlet, and the height of the tobacco monitoring unit from the ground) had higher contributions to the predicted moisture content results, which could reflect the influences of air flow movement and temperature and humidity gradients in the bulk curing barn on the spatial distribution of tobacco leaf moisture content. 3) At the yellowing stage (38 ℃), color fixing stage (45 ℃) and stem drying stage (68 ℃), the distribution ranges of the moisture content at different positions in the bulk curing barn were 5.73%, 8.59% and 10.35%, respectively; and the average relative errors between the predicted and measured leaf moisture content were 1.56%, 1.61% and 3.34%, respectively, indicating that the accuracy of prediction results was good. The established SSA-BP prediction model provides technical support for on-line monitoring of tobacco leaf moisture content in the bulk curing barn.