Metritis, a prevalent disease on dairy farms, is negatively associated with reproduction, milk production, and the welfare of cows. The objective of this study was to evaluate the efficacy of monitoring low-resilience activities (i.e., behaviors that typically decrease when energy resources are limited or when the cost involved in the activity increases; e.g., brush usage) in the early detection of metritis. Data on daily brush usage (i.e., proportion of cows using the brush and the duration of usage) were collected from 28 metritic and 60 control cows 28 d postpartum using an automated monitoring system developed for the purpose of this study. During the first week following partum (before clinical diagnosis), we found no differences in brush usage between sick and control cows. However, 8 to 21 d postpartum (the week of clinical diagnosis and the first week of medical treatment), a lower proportion of metritic cows used the brush compared with control cows (0.49 compared with 0.64, respectively, at brushes installed away from the feed bunk). In addition, the daily duration of brush usage was 50% lower among cows diagnosed with metritis compared with control cows 8 to 28 d postpartum (44 s/d compared with 88 s/d, respectively). The results of this study suggest that on-farm monitoring of low-resilience behaviors, combined with existing systems that monitor core behaviors (e.g., activity and rumination), may serve as an improved method for detecting events that compromise the welfare of animals. The slow recovery of low-resilience behaviors following medical treatment (wk 4) might serve as a particularly useful indicator of progress of recovery from disease.