Micro- and nano-fluidic devices are under development for a variety of applications including bio-molecular separation, drug delivery, biosensing and cell transplantation. Regulatory approval for the commercialization of these products requires the ability to fabricate a large number of these devices with high reproducibility and precision. Though traditional microscopy and particle rejection characterization techniques provide extremely useful measurements of nano-features, they are expensive and inadequate for quality control purposes. In this study, an agile and non-destructive selection method is presented which combines a predictive theoretical model with experimental analysis of convective nitrogen flow to detect structural defects in complex drug delivery membranes (nDS) combining both micro- and nanochanneled features. The mathematical model developed bridges the fluid dynamics between the micro- and nano-scales. An experimental analysis of gas flow was performed on a total of 250 membranes representing five different channel size configurations. The accuracy and reliability of this test in detecting major and minor defects of various kinds were verified by comparing the experimental results with the theoretical prediction.