This article details a study that investigated 55 elementary pre-service teachers' content knowledge about density. Using a mixed-methods approach, pre/post face-to-face interviews and pre/post concepts maps (Cmaps) were used as data to document changes in teacher knowledge which occurred over a 15-week science methods course. Thematic analysis was used to capture patterns within the interview data and a paired-sample t-test was conducted to compare pre/post Cmap scores. Results before instruction indicate a weak framework of prior knowledge about density which was a mosaic of alternative conceptions and a few learned concepts. Many participants focused on a single aspect of density and were unable to engage in relational causality which involves the consideration of two variables simultaneously. After instruction, robust alternative conceptions continued to be observed (e.g., density is buoyancy and density is the same as heaviness, mass, or weight) and learned concepts were concrete and rote. The most common learned concept was the density algorithm (~40%) followed by the learned concept that density is an intensive property of matter (22%) and demonstrated understandings of proportional reasoning (7%). Results of the paired sample t-test demonstrate a statistically significant difference between the total proposition accuracy scores for pre/post Cmaps (t = -3.178, p < 0.002) with the instructional intervention larger for post-Cmap scores (M = 1.02; SD = 1.063) than for pre-Cmap scores (M = 0.55; SD = 0.812). The effect size was medium.