This study investigates buckling behavior of functionally graded material (FGM) plates containing composition imperfections under uncertain conditions. The plate kinematics are described by an established hyperbolic shear deformation theory, and material gradation is represented by a power-law distribution. Composition imperfections are modeled as deviations in the through thickness ceramic volume fraction, representing ceramic-rich and metal-rich distributions. Governing equations are derived using Hamilton’s principle and solved by Navier’s method for simply supported rectangular plates under uniform biaxial compression. Uncertainties in elastic moduli, plate thickness, power-law exponent, and applied compressive load are treated as random variables. The First-Order Reliability Method (FORM) with the standard HL-RF algorithm and Monte Carlo simulation (MCS) are employed to estimate reliability index and probability of failure. A parametric study is performed to examine the effects of aspect ratio, length-to-thickness ratio, power-law exponent, and composition imperfection type. The results show that ceramic-rich composition imperfections slightly increase the critical buckling load, whereas metal-rich imperfections reduce it. Sensitivity analysis indicates that the applied compressive load is the dominant uncertainty source. The contribution is the probabilistic buckling assessment of compositionally imperfect FGM plates.