When real dynamical systems or processes are modelled using ordinary differential equations, the question arises of gauging the sensitivity of predictions derived from the model to possible perturbations in various uncertain or assumed expressions or functions which appear in the differential equations. Often, for example, various terms will be assumed to be linear, and sensitivity analysis should then study not only the effects of perturbing the constant of proportionality (parameter sensitivity analysis) but also the effect of abandoning the assumption of linearity (functional sensitivity analysis).