An updated version of the computer program EXCAD [1] allows the user to optimize experimental design to estimate parameters of particular semi-parametric models. The semi-parametric models take the general form of a function of time ( t): Y( t) = NL( c( t, α), β). The function NL( C( t, α), β), is, in general, a non-linear transformation of a function, C( t, α), that in turn is the convulution of two others. One of these two functions is expressed in a non-parametric form, and is not of direct interest to the experimenter. The other is of direct interest: it is a parametric function depending on a set of parameters α. This semi-parametric model applies to numerous kinds of biological experiments, such as pharmacokinetic/pharmacodynamic, physiological, circulatory flow experiments. This paper presents a new method for determining an optimal experimental design to estimate the parameters α and β. The new approach adopts the D-optimal criterion, and is illustrated using real thiopental data.