Underbalanced drilling (UBD) is one of the most widely used technologies preferred in depleted and/or low-pressured formations. In order to achieve underbalanced conditions, drilling fluids are usually gasified. Major drilling fluids preferred during UBD are pure gas, gas-liquid mixtures, and foams. This study is focused on gas-liquid mixtures. As the gas is introduced, the behaviour of the drilling fluid becomes hard to explain for many reasons. First of all, gas is compressible and physical properties of gas are very sensitive to changes in pressure and temperature. Second, a multiphase flow phenomenon arises. When there is multiphase flow, flow patterns should be considered. It is known that there is a difficulty to predict hydraulic behaviour of gas-liquid mixtures owing to this flow pattern dependence. During a drilling operation, one of the parameters that should be considered is hole cleaning. Hole cleaning is a challenging task even for a single-phase drilling fluid. Moreover, there is still a lack of information about how the cuttings are transported when gas-liquid mixtures are used as drilling fluids. Flow-rate optimization during UBD operations for liquid and gas phases are usually conducted based on formation pressures only. However, considering only this criterion as the optimization objective is misleading and may cause serious problems during the drilling operation. In this study, gas and liquid flow rates during UBD operations are conducted not only based on formation pressures, but also based on effective hole cleaning. It is assumed that liquid phase is the major contributor for cuttings transport, and gas phase is only influencing the bottomhole pressure. A mechanistic model is introduced for estimating the hydraulic behaviour of gas-liquid mixture drilling fluids under different flow patterns. Based on the bottomhole pressure and effective hole cleaning point of view, an algorithm is proposed for estimation of the optimum required flow rates for liquid and gas phases based on the introduced mechanistic model. Also, the model predicts the required backpressure that has to be applied.