Abstract Composites are known to display a considerable amount of scatter in their material properties due to large number of parameters involved with the manufacturing and fabrication processes. This paper is concerned with the effect of random system properties on critical thermal buckling temperature of composite laminated plates with temperature dependent properties using micromechanical approach. System parameters are assumed as independent random variables. In this analysis, based on the classical lamination theory in conjunction with the Hamilton’s principle, the basic formulation of random eigenvalue problem has been deduced. A mean-centered first order perturbation technique is used to compute the second-order statistics (mean and standard deviation) of the critical thermal buckling temperature. The performance of outlined stochastic approach has been validated by comparing the present results with those available in the literature and independent Monte Carlo simulation. The effect of random material properties, thermal expansion coefficients, fiber volume fractions, aspect ratios, laying angels and boundary conditions on the critical thermal buckling temperature are presented.