Two models for the digestion of starch in foods are considered: sequential, where two or more reactions follow one another, and parallel, where they occur simultaneously. Each reaction is assumed to be first-order, and to be characterized by a rate coefficient and a digestible fraction. For parallel kinetics, a new fitting method is developed; methods for sequential reactions are already available, based on the “logarithm of slopes” technique. Least-squares fitting methods for each are set out, which identify the uncertainties associated with each rate parameter, so that one can see if one has assumed too many or too few processes. The methods gave robust results with synthetic data, and were applied to real in vitro digestion data for wheat starches with up to 93% amylose. The combination of parallel and sequential models provides a general approach to investigating the kinetics of starch digestion in food.