Subject of this paper is ASN-Minimax (AM) double sampling plans by variables for a normally distributed quality characteristic with unknown standard deviation and two-sided specification limits. Based on the estimator p * of the fraction defective p, which is essentially the Maximum-Likelihood (ML) estimator, AM-double sampling plans are calculated by using the random variables p * 1 and p * p relating to the first and pooled samples, respectively. Given p 1 , p 2 , α, and β, no other AM-double sampling plans based on the same estimator feature a lower maximum of the average sample number (ASN) while fulfilling the classical two-point condition on the corresponding operation characteristic (OC).