Historically, earth ovens have been used to provide direct evidence of ancient plant use through the recovery of charred macrobotanical remains and indirectly by means of experimental archaeology and the ethnographic record. Experiments suggest that direct evidence of ancient starch-rich plant use can be obtained through the recovery of starch granules deposited on fire-cracked-rock (FCR) during cooking episodes even in regions where macrobotanical remains are scarcely preserved. Starch contamination, however, can enter into the archaeological record providing “background noise.” Therefore, this study analyzes the results of the Paluxy Sand Geophyte Project to determine if archaeological starch (starch that is both cultural and ancient in origin) can be differentiated from contamination using FCR recovered from heating elements in well-preserved earth ovens at Fort Hood, Texas. FCR, non-cultural rock control samples (RCS), and air control samples (ACS) were processed and analyzed from 27 earth ovens at 6 sites. Contamination control measures were used, including the use of a clean bench, powder-free latex gloves, washing samples prior to processing, spot sampling, and comparisons between starch granule assemblages recovered from FCR and control samples. Laboratory and field equipment were processed and analyzed for contamination. Only one feature (Feature 4 from 41CV984) yielded starch granules that are unambiguously archaeological in origin, rather than the result of contamination, whereas starch assemblages from the other sites could be archaeological or contamination in origin. Small sample sizes, differential preservation, and/or the cooking of non-starch-rich plants could account for the lack of differences between FCR and RCS samples. Finally, maize (Zea mays) starch granules were recovered from all sample types suggesting that maize starch, most likely from “powder-free” gloves and air-fall is a significant source of starch contamination.