The developments of fields in deep waters (5000 ft and more) is a common occurrence. It is inevitable that production systems will operate under multiphase flow conditions (simultaneous flow of gas-oil-and water possibly along with sand, hydrates, and waxes). Multiphase flow prediction tools are essential for every phase of the hydrocarbon recovery from design to operation. The recovery from deep-waters poses special challenges and requires accurate multiphase flow predictive tools for several applications including the design and diagnostics of the production systems, separation of phases in horizontal wells, and multiphase separation (topside, seabed or bottom-hole). It is very crucial to any multiphase separation technique that is employed either at topside, seabed or bottom-hole to know inlet conditions such as the flow rates, flow patterns, and volume fractions of gas, oil and water coming into the separation devices. The overall objective was to develop a unified model for gas-oil-water three-phase flow in wells, flow lines, and pipelines to predict the flow characteristics such as flow patterns, phase distributions, and pressure gradient encountered during petroleum production at different flow conditions (pipe diameter and inclination, fluid properties and flow rates). The project was conducted in two periods. In Period 1 (four years), gas-oil-water flow in pipes were investigated to understand the fundamental physical mechanisms describing the interaction between the gas-oil-water phases under flowing conditions, and a unified model was developed utilizing a novel modeling approach. A gas-oil-water pipe flow database including field and laboratory data was formed in Period 2 (one year). The database was utilized in model performance demonstration. Period 1 primarily consisted of the development of a unified model and software to predict the gas-oil-water flow, and experimental studies of the gas-oil-water project, including flow behavior description and closure relation development for different flow conditions. Modeling studies were performed in two parts, Technology Assessment and Model Development and Enhancement.