We expose a potential flaw in standard treatments of weak decay amplitudes, including that of epsilon'/epsilon. We show that (contrary to conventional wisdom) dimension-eight operators do contribute to weak amplitudes, at order G_F alpha_s and without 1/M_W^2 suppression. We demonstrate the existence of these operators with a simple weak hamiltonian and proceed to discuss their contribution in a general context. We estimate the effect of dimension-eight operators to be at the 100% level in a sum rule determination of the operator O_7 for mu = 1.5 GeV, suggesting presently available values of mu are too low to justify neglect of these effects.