Thermogravimetric analysis (TGA) has been shown to be a useful techniq
ue to determine the content of starch and other components that are us
ually present in blends of starch with synthetic polymers. However, so
me of these blends are composed of elements with an important superpos
ition in their temperature degradation range. In these cases interpret
ation of the TGA results becomes quite difficult and usually important
errors are committed in the determination of blend composition We pre
sent here a method to improve the accuracy of the interpretation of TG
A data. The analysis was developed to study blends containing starch,
an ethylene-vinyl alcohol copolymer, and water plus glycerin as plasti
cizers. Using this procedure it is possible to predict blend compositi
on with an error less than 3%, even when there is an appreciable super
position between the temperature degradation range of starch and poly(
ethylene-co-vinyl alcohol). As the proposed method of analysis is quit
e general it may be extended to be used with other types of blends tha
t give thermograms with similar characteristics to those described in
this paper.