pagesReservoir parameters from the well test data are essential for reservoir management. Especially the identification the presence of the reservoir boundaries is important for an appraisal well testing. Providing reserve estimation, identifying new well locations and well placement and avoiding dry holes are some important outcomes. However more work on precision of the input data is needed before using the calculated well test parameters. Lots of ambiguity in the results obtained from well tests should be considered because these input data is inevitably subject to estimation errors. When used in well test interpretation, each of them also brings its own source of errors.In this thesis, uncertainties caused by input parameters (rock properties and fluid properties) and measured data (flow rate and pressure) is discussed rigorously. By using the determined input parameters an experimental design is constructed. To reduce the errors and increase the confidence intervals, some remedies are used such as analyses procedure (use of deconvolution), design of well tests (longer build up times and more than one build up period after sufficient flow period). After running the cases of the experimental design, the results are used to develop a linear predictive model to conduct sensitivity analysis. A real field example is presented to illustrate such errors and the applied remedies to field application.