The ever increasing activity in the areas of mathematics and science concer
ned with reaction-diffusion equations marks both their important role in mo
deling physical phenomena in such diverse fields as biology, chemistry, met
allurgy, and combustion, and the beauty and complexity found in their solut
ions. Numerical analysis of reaction-diffusion equations has become a centr
al tool in their study because of the many barriers that exist for mathemat
ical analysis. It is exactly these situations, when we know little about th
e true solution, that are particularly needful of accuracy in numerical res
ults. Yet, these same analytic difficulties also give rise to nearly insurm
ountable barriers to accurate analytic estimation of the error of numerical
solutions. In this paper, we investigate a different approach to this prob
lem based on the computational estimation of the error of numerical solutio
ns.