While there have been a number of studies on the effect of leak site and shape on the magnitude of measured leakage through respirator face seals, there have been very few studies to identify the location and size of these leaks. In a previous study we used a method of identifying the location and shape of respirator leaks on a half-mask respirator by the deposition of a fluorescent tracer during a fit test, and testing for their association with facial dimensions. The purpose of this study was to apply that methodology to conduct multiple fit tests to determine if gender, respirator brand, repeated fit tests, and test exercises affected the location and shape of face seal leak sites. Categorical analysis found that none of these factors had a significant effect on the location and shape of leaks. General linear model analysis found some significant effects of the study factors on leaks, but facial dimensions had a greater effect, and there were significant differences between facial dimensions of subjects with a leak and those without. Significant differences in leak site distributions between this and the previous study may have been due to differences in facial dimensions and racial/ethnic composition. Twice as many diffuse leaks as point leaks were observed in both studies, indicating that slit-like leaks would be most appropriate on mannequins used in laboratory respirator leakage studies, and in respirator flow and penetration models. That the study factors had no significant effects in the categorical analysis, significant effects for facial dimensions were found in the linear analysis, and leak site distribution differences between this and our previous study may have been affected by differences in facial dimensions, indicate that, in addition to size, the shape of an individual's face may be an important determinant of leak sites on a half-mask respirator. This would have implications for the design of respirator facepieces and in the selection of respirators for individual wearers.