Variation in nasal floor topography has generated both neontological and paleontological interest. Three categories of nasal floor shape (Franciscus: J Hum Evol 44 (2003) 699-727) have been used when analyzing this trait in extant humans and fossil Homo: flat, sloped, and depressed (or "bi-level"). Variation in the frequency of these configurations within and among extant and fossil humans has been well-documented (Franciscus: J Hum Evol 44 (2003) 699-727; Wu et al.: Anthropol Sci 120 (2012) 217-226). However, variation in this trait in Homo has been observed primarily in adults, with comparatively small subadult sample sizes and/or large age gradients that may not sufficiently track key ontogenetic changes. In this study, we investigate the ontogeny of nasal floor shape in a relatively large cross-sectional age sample of extant humans (n = 382) ranging from 4.0 months fetal to 21 years post-natal. Results indicate that no fetal or young infant individuals possess a depressed nasal floor, and that a depressed nasal floor, when present (ca. 21% of the sample), does not occur until 3.0 years postnatal. A canonical variates analysis of maxillary shape revealed that individuals with depressed nasal floors were also characterized by relatively taller anterior alveolar regions. This suggests that palate remodeling at about 3.0-3.5 years after birth, under the influence of tooth development, strongly influences nasal floor variation, and that various aspects of dental development, including larger crown/root size, may contribute to the development of a depressed nasal floor. These results in extant humans may help explain the high frequency of this trait found in Neandertal and other archaic Homo maxillae.