The dynamics of contact, stress and failure analysis of multibody systems is highly nonlinear. Nowadays, several commercial and other analysis software dedicated for this purpose are available. However, these codes do not consider the uncertainty involved in loading, design, and assembly parameters. One of these systems with a combined high nonlinearity and uncertainty of parameters is the gearbox of wind turbines (WTs). Wind turbine gearboxes (WTG) are subjected to variable torsional and nontorsional loads. In addition, the manufacturing and assembly process of these devices results in uncertainty of the design parameters of the system. These gearboxes are reported to fail in their early life of operation, within three to seven years as opposed to the expected twenty years of operation. Their downtime and maintenance process is the most costly of any failure of subassembly of WTs. The objective of this work is to perform a probabilistic multibody dynamic analysis (PMBDA) of a helical compound planetary stage of a selected wind turbine gearbox that considers ten random variables: two loading (the rotor speed, generator side torque), and eight design parameters. The reliability or probabilities of failure of each gear and probabilistic sensitivities of the input variables toward two performance functions have been measured and conclusions have been drawn. The results revealed that PMBDA has demonstrated a new approach of gear system design beyond a traditional deterministic approach. The method demonstrated the components' reliability or probability of failure and sensitivity results that will be used as a tool for designers to make sound decisions.