The cost of feed is the largest expense of a dairy farm, and this cost is rising. Feed intake traits such as DMI are highly desirable traits for dairy breeding programs. Measuring individual feed intake, however, is difficult and expensive. The use of predictor traits may be better suited to measure feed intake, as they are less expensive and easier to measure. Information such as mid-infrared spectroscopy (MIR), BW, and BCS have the potential to be predictor traits for DMI. MIR is used to analyze molecular vibration and rotation when a material is exposed to electromagnetic radiation. It is currently used worldwide to quantify milk components during routine milk analysis. MIR technology may provide a cost-effective opportunity to obtain predicted phenotypes for feed intake on many animals by taking advantage of technology that is currently used in regular milk recording. The objective of this study was to evaluate the efficacy of MIR, BW, and BCS to predict DMI and to determine the optimal predictor trait for DMI. Weekly milk samples were collected for 143 Canadian Holsteins (n = 2,775) during routine milk recording and were sent to CanWest DHI for MIR spectral analyses. Milk, fat and protein yields (n = 2,775), BW (n = 1,656), and BCS (n = 1,656) were also collected. Daily DMI (kg/d) for all cows (n = 2,775) was recorded and averaged per lactation week with a mean (SD) of 85.33 (43.99). A prediction equation for DMI was produced using partial least squares regression from the MIR spectra of milk samples. Equations to predict DMI using BW and BCS were calculated to determine the optimal predictor trait for DMI. Further steps are needed to determine the ability of the prediction equations to estimate DMI.