Reducing antimicrobial use in dairy cattle requires practical methods to identify cows that need antimicrobial treatment at dry-off. The objective of this study was to develop and evaluate a classification model using on-farm indirect indicators as a first step toward identifying cows with intramammary infection (IMI) at dry-off. The study was conducted on a single commercial dairy farm and included 301 cows. Quarter milk samples were collected before dry-off and used as the reference method to define IMI. Cow-level predictors included somatic cell count (SCC) history, clinical mastitis history, parity, teat health score, and genetic values for SCC, which were combined in a multivariable logistic regression model internally validated using a 75/25 training-testing split. Overall, 1190 quarters were examined, and 129 cows were classified as infected with at least one major mammary pathogen at dry-off. The best-performing model included genetic value for SCC, linear score at selected test days, and clinical mastitis history, reaching an accuracy of 0.75 (0.64–0.84), a sensitivity of 0.80 (0.63–0.92), a specificity of 0.71 (0.55–0.84), and an AUC of 0.76 (0.65–0.87). These results suggest that combining indirect cow-level indicators may support treatment allocation at dry-off, although external validation across multiple herds is required.