In this paper, a new sampling plan is introduced based on the run length of conforming items and Taguchi loss function. Run length of conforming items is used for decision making about the quality of a lot. The proposed sampling plan is based on the Markov modeling. A continuous Taguchi loss function is used to obtain loss of deviations between the value of quality characteristics and its target. An optimization model is developed for obtaining the optimal control tolerances and the corresponding critical acceptance and rejection thresholds based on the geometric distribution, which minimizes the loss function for both producers and consumers. For practical purpose, a numerical example is presented to illustrate how the proposed procedure can be applied to design acceptance sampling plans. In addition, the results of a comparison study denoted the merits of proposed model. At the end, a general model regarding the risk of consumer and producer is developed. A sensitivity analysis is carried out on the different values of the required parameters in order to validate the results of optimization model.