Melt index (MI) is universally accepted as the primary specification in the thermoplastics industry. The standard according to which the MI measurement is made is ASTM D-1238. However, some variation in the MI measurement is possible, even when all ASTM D-1238 specifications are met. A thorough understanding of the sources and magnitude of variation is essential, since measurement error can have significant economic implications. Error in the MI measurement can cause a resin lot (about 190,000 pounds) to be erroneously classified as off-spec (leading to a loss of over US $20,000). A Resolution IV, six-factor, 1/4 fractional factorial design of experiments (DOE) study [with three replicates] was done to quantify the effect of various factors affecting variability in the MI measurement using two high-density polyethylene monitor resins with MI of 1.91 g/10 min("low" MI) and 36.4 g/10 min("high" MI). A major challenge was dealing with qualitative (e.g., "cleanliness") and quantitative (e.g., "sample mass" or "temperature") factors in the same DOE. For the high MI monitor, among the factors considered, the most significant (for the ranges considered) were found to be (in the order of importance) : die orifice diameter, temperature, die cleanliness, barrel cleanliness, and sample mass. The following factors were also considered but not found (in the ranges considered) to be as significant : piston land (tip) diameter, load, piston cleanliness, preheat time, and piston curvature. For the low MI monitor, among the factors considered, the two most significant (for the ranges considered) were temperature and barrel cleanliness. This article gives practical and useful information for those who make and/or use MI measurements. As others have pointed out, much of such discussion related to melt indexers is also applicable to more sophisticated capillary rheometers.