Reduced mechanisms are often used in place of detailed chemistry because the computational burden of including all the species continuity equations in the reactor model is unreasonably high. Contemporary reduction techniques produce mechanisms that depend strongly on the nominal set of problem parameters for which the reduction is carried out. Effects of variability in these parameters on the reduced mechanism are the focus of this work. The range of validity of a reduced mechanism is determined for variations in initial conditions. Both sampling approaches and quantitative measures of feasibility, such as the flexibility index and the convex hull formulation, are employed. The inverse problem of designing a reduced mechanism that covers the desired range of initial conditions is addressed using a multiperiod approach. The effect of the value of a user-defined tolerance parameter, which determines whether the predictions made by the reduced mechanism are acceptable, is also assessed. The analytical techniques are illustrated with examples from the literature.