Currently, construction cost overrun is a huge challenge not only in developing countries like Ethiopia but also in developed countries as well. One of the reasons is lack of proper planning and preliminary cost estimation as the scope is not fully developed and there is limited information at the early stages, which may have a lot of uncertainties. This study investigates the utilization of a relatively new cost forecasting method called reference class forecasting (RCF) founded by Israeli-American psychologist and economists called Kahneman and Tversky and then developed by a scientist Bent Flyvbjerg. The study determines the extent of cost variations and assesses the feasibility of RCF in estimating costs by developing statistical distributions based on previously constructed road construction projects which can later helps to quantify the level of contingency and cost overrun risk level for the project at hand in to be appraised. The paper also explored theoretical aspect of main group factors to cost overrun expressed interms of optimism bias and strategic misrepresentations which led to cost underestimations are a critical input to conduct RCF in practice. Both quantitative and qualitative research approaches have been applied, whereby interviews, surveys and case studies are used as primary data sources. Extensive literature review has been carried out as a secondary data sources. The analysis of the research had been supported by using causal and descriptive discussions of interview; and descriptive statistics for survey data using analysis software, statistical simulation and distributions for case study projects. Microsoft Excel and Origin 8 for the case projects and SPSS and Smartpls tools were used in questionnaire analysis. Generally the finding shows that experts gave more emphasis to technical issues for project cost performance problems than to political and optimism bias effects though cost underestimation is a serious cause to project cost overrun.The extent of average cost overrun in the case project is lower than the expectations which are a reflection of outliers resulted from small sample size and the attitude of officials not to report negatives. Within these constraints the Uplift Model was developed significant for making decision in project appraisal stage besides to other commonly forecasting methods to quantify contingency with the expected risk level and then Uplift value comparison had been done with othe Countries‟ Uplift value results.