Effect of sample size (aspect ratio or ratio of thickness to lateral dimension) on the value of the apparent direct longitudinal d(33) coefficient is investigated in hard and soft PZT using quasistatic Berlincourt-type technique. We show that at small aspect ratio (such as thickness/diameter=0.1) the measured d(33) is as much as 30% lower than the true value, d(33true). Measured d(33) increases with increasing aspect ratio and reaches its true value at a threshold aspect ratio that is equal to approximate to0.5 for both soft and hard PZT. The experimental results are compared with simulations using Finite Element Modeling (FEM) and are interpreted in terms of inhomogeneous distribution of strain/stress within the sample, which leads to functional dependence of the measured d(33) on transverse d(31) and shear d(15) coefficients. Nonlinarity of the d(31) and d(15) coefficients also was taken into account in the finite element analysis. FEA and experimental results are in a good agreement.