To reduce the amount of organic solvents in pendimethalin emulsifiable concentrate (EC), small-size microcapsules (S-MCs) and large-size microcapsules (L-MCs) were prepared with polyurea as a wall material. Petri-dish bioassays were carried out to investigate the bioactivity of formulations and the influence of both organic matter and moisture. The relationships between degradation and the biological activity of three pendimethalin formulations in the soil were investigated, and field experiments were executed to verify the laboratory results. The laboratory tests showed the following: (1) the bioactivity of EC and S-MCs was similar and greater than that of L-MCs; (2) organic matter could reduce the bioactivity of MCs and EC, and the impact of organic matter on L-MCs was greater; (3) increased soil moisture content had no significant effect on the bioactivity of EC but slightly reduced that of the MCs; and (4) the L-MCs showed significantly more prolonged residual and effective persistence in the soil than did EC and S-MCs. However, the field experiments indicated that the herbicidal efficacies of L-MCs at the early and late stages were both lower than those of EC. Comprehensive analysis of the results indicated that the main reason that the herbicidal efficacy of L-MCs was lower than that of EC in the field was that L-MCs missed the optimal herbicidal periods due to the slow-release characteristics of L-MCs. The S-MCs had both similar release rates and herbicidal efficacy in the field as EC. Therefore, to develop a good pesticide formulation, the occurrence and damage characteristic of pests must be considered.