A standard method for evaluating weed seed predation is needed to facilitate generalizations across studies. Identification of general trends could allow practical recommendations for enhancing weed seed predation in agricultural systems. The objective of this study was to compare the commonly used sandpaper and soil substrate methods for offering weed seeds when assessing seed predation rates. Invertebrate seed predators and associated weed seed predation levels were measured in June to July, August, and September of 2005 and 2006 within a conventionally managed corn system. Seed predation levels of three common weed species, velvetleaf, giant foxtail, and common lambsquarters, were estimated using feeding trials in which 40 seeds of each species were offered over a 48 h period using the two substrates. Exclosures were used to distinguish total predation from predation by invertebrates alone. In addition, we investigated the use of geospatial analysis to estimate spatial autocorrelation of invertebrate populations and seed removal rates. Results suggest caution in using synthetic substrates, such as sandpaper, when assessing seed predation, especially when investigating small-seeded species (< 1 mg seed−1) or when seed predators are predominantly invertebrates. By contrast, predation of the larger-seeded species, velvetleaf, was less affected by substrate, perhaps because of removal predominately by vertebrates. One way to overcome problems with the sandpaper substrate method is for studies to include some soil substrate samples for on-site calibration of the sandpaper substrate. If necessary, data could then be corrected by multiplying by the ratio of soil substrate measured-predation rate to sandpaper measured-predation rate. Spatial autocorrelation explained between 6 and 9% of the variation in giant foxtail and common lambsquarters removal rates attributed to invertebrates alone. Researchers should, therefore, be careful not to neglect the impact of clustered invertebrate populations and associated seed removal rates.