A strategy is presented for scrutiny of the quasi-steady-state (QSS) approximation in metabolic modeling that involves tracking dynamic data on extracellular products in so-called flux, as well as yield vector spaces when the substrate uptake flux is perturbed sinusoidally about some average value with specified frequency and amplitude. Criteria accrue for the diagnosis of any violation (and its extent) of the QSS approximation and are applied to data produced by simulations from models free from the QSS approximation. It is shown that, even when the assumption is admissible, the choice of the appropriate QSS model depends on the frequency and amplitude of the perturbation. Three different QSS models in the literature, viz., dynamic flux balance analysis, the macroscopic bioreaction model, and the hybrid cybernetic model, are evaluated, and circumstances are identified when one or more of the QSS models might be valid, as well as when none of them are.