Observation of archaeological charcoal records from Maya sites, and in particular, the Classic site of Naachtun (Northern Peten, Guatemala), indicates that certain woody taxa tend to occur in proportions that are highly inconsistent with their representation in the local forests today. We note this phenomenon for two taxa in particular: 1) the genus Manilkara, which dominates the charcoal assemblages of Naachtun, but grows in relatively low proportions in modern Central Lowlands forests, and 2) breadnut (Brosimum alicastrum), which is widespread in the modern forests of the region, but whose wood is almost absent in the archaeological record. Based on ethno-graphic and ethnohistoric accounts, many researchers have argued that both of these trees would have played a major role in ancient Maya agroforestry. Therefore, it becomes necessary to determine how accurately the occurrence of Manilkara and Brosimum in archaeological charcoal records reflects their use in the past. We explore the hypothesis that combustion processes may create taphonomic biases that lead to the differential preservation of certain Maya Lowland tree taxa, and thus distort the representivity of the charcoal spectra recovered from ancient Maya sites. To evaluate this hypothesis, we conducted 35 experimental fires using five of the principal tree species of the modern forest around the site of Naachtun, including Manilkara and Brosimum. The charcoal assemblages produced through these fires were systematically identified and studied using quantitative methods. Our results indicate that significant differences exist among these taxa as a result of combustion, and that these phenomena are consistently observed for each taxon through multiple controlled trials. Thus, anthracological analyses are indeed appropriate for reconstructing human-environmental interactions in the Central Lowland forest, but certain predictable taphonomic biases must be taken into account when interpreting the charcoal data.