Geometric mean residence time was compared with the arithmetic mean residence time for a residence time distribution study conducted on a twin-screw extruder using a salt tracer and electrical conductivity measurements. Electrical conductivity of the cornmeal melt was measured at the die exit. All RTD curves obtained during the study were skewed to the right with very long tails that resulted in inflated arithmetic mean residence times. In order to alleviate the problem of long tails, different approaches were considered. Among these approaches, logarithmic transformation of the time scale to calculate the geometric mean residence time was shown to best represent the residence time distribution in terms of mean residence time and spread of the RTD. The log transformation of the time scale alleviates the problem of inappropriate shifting of the mean residence time due to long tails. The difference between the upper limits and lower limits for the log transformed data was found to be more appropriate to define the spreads of the distributions as the untransformed data resulted in quite variable spreads.