Posture prediction is a key component in digital human modeling and simulation. Deterministic optimization-based posture prediction formulations have been proposed. However, there exist uncertainties in human anthropometric (human height, link length, center of mass of segments, and moment of inertia, etc.) and environment parameters (location, interaction force), which affects the predicted posture. This paper attempts to study the effect of uncertainty on predicted posture. The single-loop reliability based design optimization (RBDO) method is adapted to predict posture under uncertainties. All random parameters are assumed to have normal distribution. A 24-degree of freedom (DOF) upper body model is used. In this pilot study, it is assumed that one link length and one joint angle are random parameters. The other design variables and parameters are deterministic. With the empirical rule, three cases are investigated for posture prediction. SNOPT software solver is employed to solve the optimization problem. Through comparison with deterministic optimization result and experimental data, the predicted postures from RBDO simulation show that the reliability index affects the predicted posture to some extent.