Between 1990 and 1994, 18 horizontal wells were drilled and completed in the Pembina Cardium pool. Increased application of this technology in the pool has been curtailed by limited understanding of the performance of these wells and lack of reliable prediction models. This paper focuses on the analysis of performance of the 18 wells and the development of a methodology leading to a reliable performance prediction model for horizontal wells in the pool. The model was developed using analytical modelling techniques. A major advantage of the analytical technique over reservoir simulation, besides its low cost and quick turn-around, is that it involved more focus on key performance drivers. The analytical model couples a material balance model with a reliable horizontal well productivity model. This methodology provides deep insights, particularly into the performance of horizontal wells within a stratified reservoir under waterflood. It shows that traditional methods of modelling horizontal wells by history matching vertical wells or using type curves to establish reservoir parameters may lead to inaccurate or unreliable performance predictions, particularly in layered reservoirs under waterflood. A material balance model coupled to a proper horizontal well productivity model, the application of a detailed reservoir description, and correct averaging techniques are critical to successful performance forecasting. The methodology developed in this study has resulted in the identification of key reservoir characteristics leading to selection of reservoir targets for successful application of horizontal wells in the Pembina Cardium pool. Introduction The Pembina Cardium pool, located 120 km southwest of Edmonton, is the largest conventional oil pool in Canada, containing 1.1 billion standard m3 of original oil in place(1). The reservoir is a stratigraphic trap and has neither an aquifer nor a gas cap. The discovery well was drilled in 1953. The Cardium formation is extremely heterogeneous and is composed of low- to medium-permeability sandstone members which in places are overlain by a highly permeable conglomerate unit. Seven parasequences, separated by impermeable shale units, comprise the sandstone. A parasequence is a relatively conformable succession of genetically related beds or bedsets bounded by marine-flooding surfaces or their correlative surfaces. The main recovery method is waterflooding(2–5) and about one third of the 5,200 wells in the pool are water injectors. The pool is currently produced by pattern waterflooding using inverted 5spots, 9-spots, and line drives. The reservoir stratification and large differences in permeability between layers have caused the waterflood to be inefficient in sweeping the lower permeability layers. Oil recovery to date in the Pembina Cardium pool is estimated at 23% of OOIP. In late 1993, the eight largest operators of the Pembina Cardium pool created the Pembina Cardium Renewal Consortium (PCRC). The PCRC was created to determine how to increase both recovery and recovery rates in the pool. It was staffed by engineers and geologists from four of the member companies, and contract consultants. The Alberta Department of Energy, the Energy Resources Conservation Board (ERCB), and internal and commercial laboratories were involved as needed.