Abstract Wood density is used for genetic selection because of its association with wood quality. Traditional assessments of wood density, such as volumetric measurement and X-ray densitometry, are labor intensive and time consuming. Drill resistance (measured with a Resistograph) is faster and gives an indirect assessment of wood density. The Resistograph generates thousands of data points for each tree that must be summarized. The data include bark and air as the drill enters and exits the bole, as well as friction that builds on the drill shaft. This study evaluated data processing methods for drill resistance to improve correlation with wood density, with a focus on practical implementation by breeding programs. We compared methods that thoroughly removed bark, air, and friction, as well as simple methods based on expected bark thickness and depth to the pith. The most thorough method produced the highest phenotypic correlation with density ( r p = 0.81). All methods that excluded data after the drill exits the tree had very strong genetic correlations ( r g > 0.92), although avoiding bark gave a noticeable increase in genetic correlation. Unprocessed drill resistance had unacceptably low phenotypic correlations ( r p = 0.38) and the lowest genetic correlation observed, although it was still very high ( r g = 0.91). For tree breeding programs, we recommend developing reasonable bark and pith depth thresholds and using simple rules to avoid bark and friction.