The actual transverse and longitudinal displacement of PZT thick film was measured using a newly developed atomic force microscopy (AFM). The AFM is attached a feedback circuit named "torsion feedback". The torsion and Z-height feedback circuits control an AFM cantilever to follow piezoelectric deformation of the sample. To measure transverse displacement, the cantilever contacts the edge of sample. The transverse displacement is determined from the torsion feedback signal absolutely. To measure longitudinal displacement, the cantilever contacts the center of sample. The longitudinal displacement is determined from Z-height feedback signal absolutely. A 5-μm-thick PZT film was prepared on Pt/Ti/SiO 2/Si substrates. The film sample was shaped square pillar. The side electrode length (L) of square pillar shaped sample was ranged from 1000 μm to 10 μm. The relation between side electrode length and the transverse or the longitudinal displacements were investigated. With decreasing L, the transverse displacement decreased nonlinearly, and the longitudinal displacement increased nonlinearly. The finite element method (FEM) simulation suggests that the substrate clamped PZT film behaved nonlinearly. The effective -d 31 and d 33 were calculated from the measured displacement, and these values increase with decreasing L. The effective d 33 and -d 31 showed correlation, and the ratio was d 33:-d 311 = 5.3:1, whereas the bulk ratio is d 33:-d 31 = 2.4:1. This result suggests that the substrate clamping effect of the transverse displacement was larger than that of the longitudinal displacement.