Lameness in dairy cows is a concern for both producers and consumers. Milk mid-infrared (MIR) analysis could be an extra tool in the detection of lameness problems for farmers. The aim of this study was to test the feasibility of detecting lameness problems using MIR spectra from milk through the development of predictive models. The data for this research was provided by RINDERZUCHT AUSTRIA (2017), from their “Efficient Cow” project and were recorded between July 2014 and December 2014. The data sets used were the complete data set of 9811 records and subsets according to lactation stage, parity, breed and hoof disease. Two types of pre-processing were tried: first derivative followed by a Standard Normal Variate (SNV) transformation or second derivative followed by a SNV transformation. The first and second derivatives do not give the same results which highlight the importance of pre-processing during model development. The best results were obtained for the Heel horn erosion subset. However, the specific nature of the used data requires the addition of more data coming from varied animals and farms and validation steps before using this technology on a larger scale.