Many techniques are available to assess masticatory performance, but not all are appropriate for every population. A proxy suitable for elderly persons suffering from dementia was lacking, and a two-colour chewing gum mixing ability test was investigated for this purpose. A fully automated digital analysis algorithm was applied to a mixing ability test using two-coloured gum samples in a stepwise increased number of chewing cycles protocol (Experiment 1: n = 14; seven men, 19-63 years), a test-retest assessment (Experiment 2: n = 10; four men, 20-49 years) and compared to an established wax cubes mixing ability test (Experiment 3: n = 13; 0 men, 21-31 years). Data were analysed with repeated measures anova (Experiment 1), the calculation of the intraclass correlation coefficient (ICC; Experiment 2) and Spearman's rho correlation coefficient (Experiment 3). The method was sensitive to increasing numbers of chewing cycles (F5,65 = 57·270, P = 0·000) and reliable in the test-retest (ICC value of 0·714, P = 0·004). There was no significant correlation between the two-coloured gum test and the wax cubes test. The two-coloured gum mixing ability test was able to adequately assess masticatory function and is recommended for use in a population of elderly persons with dementia.