A degree of difference test and its statistical analysis were developed to replace the traditional triangle test for products which vary during production. Several heterogeneous food products and computer simulated data were used in this study. The result shows that the use of the triangle test with a highly variable product results in a high frequency of false statistically significant results being declared. A degree of difference test which measures and tests perceived differences in taste due to other than product variability via the analysis of variance approach is therefore recommended for dealing with heterogeneous products.