Collagen associated with bone samples is frequently used for radiocarbon (14C) dating of bones recovered from archaeological sites. However, submersion and exposure to moisture favors the degradation of collagen, which leads to difficulty in reliably dating bones from tropical, humid, or previously submerged archaeological sites. In this paper, we characterized the preservation state of a series of bones, through parameters such as %C, %N, C/N ratio, and collagen recovery. We performed 14C analyses of three collagen fractions obtained through the pretreatment steps (total, ultrafiltered, and insoluble collagen) in order to link the preservation state and the reproducibility of 14C values obtained from the three fractions. Collagen ultrafiltration resulted in a decrease of C/N ratio, although collagen yield was reduced. When two or three collagen fractions were obtained, ages were reproducible and consistent with expected values, according to archaeological or hydrogeological criteria. The pretreatment steps were monitored by infrared spectroscopy in order to analyze the collagen fractions at the molecular level. The presence of collagen in the total and insoluble fractions was confirmed. Since many of the Mexican samples had poor ultrafiltered collagen yield (<3%) or nonexistent yield, our results show that if additional contextual information is carefully considered, the remnant collagen in the total and insoluble fraction can be dated, especially from sites where no other datable fraction exists.