Rapidly prototyping a polymer microfluidic device is a growing interest for various applications and micromilling is one of the common approaches to manufacture polymer microfluidic devices. The advantages of using micromilling for polymer microfluidic devices include shorter fabrication process, lower cost, easier user interface, and being capable of fabricating complicated structures, which make micromilling a perfect tool in research idea testing and validation. The aim of this study is to understand the impact of each micromilling parameter to the surface roughness on a polycarbonate substrate, followed by the factor analysis to determine the optimal cutting conditions. The parameters included spindle speed, feed rate, and the depth of cut, and the roughness was measured by a stylus profilemeter. The smallest roughness achieved was $0.127{mu}m$ with the spindle speed of 20,000 rpm, feed rate of 300 mm/min, and the depth of cut of $10{mu}m$. Factor analysis was used to determine that the spindle speed has the largest impact while the depth of cut has the minimized impact to the surface quality. To further confirm the optimal cutting conditions, twelve reservoirs were micromilled with the optimal cutting conditions and the average roughness is $0.14{mu}m$ with a stand deviation of $0.08{mu}m$.