In this investigation, the potency distribution of a low-dose drug in a granulation was optimized through a two-part study using statistically designed experiments. The purpose of this investigation was to minimize the segregation potential by improving content uniformity across the granule particle size distribution, thereby improving content uniformity in the tablet. Initial operating parameters on the Gerteis 3-W-Polygran 250/100/3 Roller Compactor resulted in a U-shaped potency function (potency vs. granule particle size) with superpotent fines and large granules. The roller compaction optimization study was carried out in two parts. Study I used a full factorial design with roll force (RF) and average gap width (GW) as independent variables and Study II used a D-optimal response surface design with four factors: RF, GW, granulating sieve size (SS), and granulator speed (GS). The planned response variables for Study I were bypass weight % and potency of bypass. Response variables for Study II included mean granulation potency with % relative standard deviation (% RSD), granulation particle size, sieve cut potency % RSD, tablet potency with % RSD, compression force at 7 kP crushing strength, and friability of 7-kP tablets. A constraint on GW was determined in Study I by statistical analysis. Bypass and observations of ribbon splitting were minimized when GW was less than 2.6 mm. In Study II, granulation potency, granulation uniformity, and sieve cut uniformity were optimized when the SS was 0.8 mm. Higher RF during dry granulation produced better sieve cut uniformity and tablets with improved uniformity throughout the run, as measured by stratified tablet samples taken during compression and assayed for potency. The recommended optimum roller compaction and milling operating parameters that simultaneously met all constraints were RF = 9 kN, GW = 2.3 mm, SS = 0.8 mm, and GS = 50 rpm. These parameters became the operating parameter set points during a model confirmation trial. The results from the confirmation trial proved that the new roller compaction and milling conditions reduced the potential for segregation by minimizing the granulation potency variability as a function of particle size as expressed by sieve cut potency % RSD, and thus improved content uniformity of stratified tablet samples.