Over the past few decades, wireless sensor networks have been widely used in the field of structure health monitoring of civil, mechanical, and aerospace systems. Currently, most wireless sensor networks are battery powered and it is costly and unsustainable for maintenance because of the requirement for frequent battery replacements. As an attempt to address such issue, this paper theoretically and experimentally studies a compression-based piezoelectric energy harvester, which is suitable for the low-frequency and large-force working environments, such as in civil and transportation infrastructure applications. The proposed energy harvester employs the piezoelectric structure constructed in multilayer stack configuration to convert ambient vibrations into electrical energy. Based on the linear theory of piezoelectricity, the two-degree-of-freedom electromechanical models of the proposed energy harvester were developed to characterize its performance in generating electrical energy under external excitations. Exact closed-form expressions of the electromechanical models have been derived to analyze the maximum harvested power and the optimal resistance. The theoretical analyses were validated through several experiments for a test prototype under harmonic excitations. The test results exhibit very good agreement with the analytical analyses and numerical simulations for a range of resistive loads and input excitation levels.