Abstract: A new content uniformity testing scheme by variables for pharmaceutical preparation has been formulated in two steps: 1. The locus of the point (μ,σ) representing pairing of the two population variables of active ingredient in individual dosage units was found by statistical probability computation and approximated to two connecting lines, one circular segment and one straight line, the equations of which contain the two sample variables (?) and S; 2. Two straight lines delimiting a double sampling zone were found by weighing the two contradictory indices, high accuracy and small sample size of the sampling inspection scheme. The sample size in the initial inspection is n₁=10, while that in the second inspection is n=n₁+n₂=30. The 4 discriminants of the new scheme are as follows:A₁+k₁S₁=l, A₁ +k₂S₁ =lA₂+S₂=k₃, A+k₄S=l where l is the tolerant content limit of a dosage unit; k₁, k₂, k₃ and k₄ are parameters that depend on l; and other symbols are self--evident. The numbers of k₁,k₂, k₃, k₄ for±l%=±15% and d%≈8, 7, 6 and 5% are given, where d is the acceptance defective in lot.The algorithms used are outlined. Some of the merits of the new scheme——accuracy, adapta-bility and practicability——are discussed.