Polymeric substrates have significant advantages over silicon and glass for use in microfluidics. However, before polymer microfluidic devices can be mass produced, it must be shown that the manufacturing method used to create these devices is robust and repeatable. For this paper, a polymer manufacturing process, hot embossing, was used to produce microsized features in polymethylmethacrylate (PMMA) chips. A design of experiments that varied two factors during the hot embossing process (temperature and pressure), was conducted to determine the robustness of hot embossing microsized channels in PMMA. The channel height and width were measured at three sites on each chip, and the results were analyzed in two ways: response surface modeling (RSM) and nested variance analysis. For the RSM analysis, two separate ANOVA tests and regressions were performed on both channel width and channel height to obtain the response surface models between temperature, pressure and the channel width and height. Furthermore, the variance of channel width and height at each design point was determined and then two ANOVA tests and two separate regressions were performed to obtain the response surface models between temperature, pressure and the variance of channel height and channel width. This analysis was used to determine if hot embossing microfluidic devices is a robust process capable of producing quality parts at different operating conditions. The nested variance analysis was used to determine the primary source of the variation in channel height and width. For the nested variance analysis, two separate calculations were performed in order to determine whether the variance of channel width and height is mostly caused by within-chip variance or chip-to-chip variance. The analysis showed that the channel widths and heights were statistically equal across the four different operating points used (the low-temperature, low-pressure point was omitted). The variance of channel width and the variance of channel height remained constant in the desired operating region. Based on this analysis, it was concluded that hot embossing is a robust process for features on the order of 50 [micro]m. Furthermore, the nested variance analysis showed that the variance of channel width and height is mostly caused by site-to-site measurements on a chip rather than between-chip variance. Therefore, it was determined that hot embossing microfluidic devices are repeatable and consistent from chip-to-chip. [DOI: 10.1115/1.4001421] Keywords: hot embossing, microfluidics, process robustness, polymer manufacturing