Some contributions of the pattern recognition methodology to design of sampling plans are described and illustrated with no prior information about the quality of lots. Specifically speaking, the modified Min-Max and Neyman-Pearson diagonosis rules are presented for design of single sampling inspection by asttributes. The recommended procedures enable us to find the following two plans. The first insures the specified consumer's risk, minimizes the maximum of the average loss due to misclassification and reduces the sampling cost. The second minimizes the sum of the agerage loss and sampling cost and insures the consumer's risk.