Geared systems are critical to the power generation and transportation industry. While thermal and multiphase flow physics are among the leading determinants of the performance, durability and life of these systems, limited capability has been developed to predict their thermo-fluid behavior. This is, in part, due to the significant complexity and multi-scale nature of the physical phenomena involved. This paper aims to address this issue and presents a novel modeling approach and supporting theoretical analysis for predicting the unsteady thermo-fluids of geared systems that experience periods of constant gear speed operation. A regime map is developed that characterizes the dominant heat generation mechanisms of geared systems. The regime map, along with a scaling analysis for the transient processes within the system is employed to quantify the separation of time-scales among various thermo-fluid phenomena. This leads to illustrate the impracticality of the conventional time marching methods of solving conservation equations to resolve the thermo-fluids of the system for the prevalent range of operating conditions. A set of numerical and mathematical approximations are established to reduce this separation based on physical motivations. In addition, a novel solution approach and associated mathematical derivation is presented that exploits the separation of time-scales and solves for the time-dependent stationary state of the system with modest computational cost. This comes at the expense of not capturing long term transient heat transfer phenomena such as the initial system warm up. The numerical approach is successfully verified against a traditional simulation approach for an artificial gear system specially constructed such that it is amenable to simulation using both approaches. Finally, the numerical approach is demonstrated on a realistic lubricated gear system. Results verify key assumptions associated with the approach and to enable the analysis of thermo-fluid physics of the system under various operating conditions. (c) 2014 Elsevier Ltd. All rights reserved.