The stability of emulsion films of the water-oil-water (W/O) type against stepwise application of DC voltage ramps is explored. Films are formed from toluene-diluted bitumen solutions and DC polarization is applied on equilibrium films having a constant diameter. Film thickness, h, is measured microinterferometrically and at a given critical voltage, Vcr, film rupture is detected. The goal is to investigate the applicability of Vcr parameter for characterization of film stability as a function of concentration of real bitumen solutions, Cs. The equilibrium film thickness, h, corresponding to each bitumen concentration, is also monitored. A description statistics, analysis of variances (ANOVA), t-tests and multiple comparison tests are employed to analyse the experimental results. The standard deviations (relative to the mean) of Vcr and h obtained are rather large, whereas the standard errors are small indicating that obtained Vcr and h data are representative to the entire film population at a specific concentration. The results for the 95% confidence intervals of means are indicative for the sensitivity of the critical voltages towards changes in bitumen concentration and suggest overall picture of increased critical voltages with the concentration. However, in terms of thickness values, confidence intervals suggest sensitivity of this parameter in relation with concentration changes but no clear trend is revealed. ANOVA shows that for both variables, Vcr and h, the overall differences in the mean values among the concentration groups are greater than would be expected by chance, i.e. there are statistically significant differences. This statistic also allows estimation of the overall impact of these effects and large ones were found. Implementation of independent t-tests and multiple comparison tests reveal in greater intimacy which concentration pairs express statistically significant differences in terms of Vcr and h as well as the strength of these effects. Multiple linear regression model was built in order to distinguish the importance of predictor variables (Cs and h) in determination of the outcome variable (Vcr). Measured critical voltages are found to be with relatively large standard deviations. It is found that the inclusion in the model of the variable thickness explains larger amount of the variation in Vcr values in comparison with the model, containing only the concentration. Thus, film thickness appears to be an important parameter that leads to improvement of the predictive power of the model. The B-values for both predictors are positive indicating positive relationships with Vcr. Therefore, as bitumen concentration increases, critical voltage increases; as film thickness increases so does critical voltage. It is inferred that the concentration has a bit larger comparable impact on critical voltage than the thickness.