Monitoring physical activity of cattle is an estrus detection tool utilized on dairy farms. Lower than normal dairy cow activity can enhance postpartum metabolic disease risk (e.g., ketosis). Considering that NEFA mobilization and BHBA synthesis can promote metabolic disease, our objective was to investigate the relationship between activity and metabolic health markers in peripartal dairy cows. Seventy-one multiparous Holstein cows (713 ± 65 kg BW) housed in a commercial herd setting were continuously enrolled. Collar mounted activity meters (DeLaval) were worn by cows −45 d before expected calving until 10 DIM. Data were recorded at 1 h intervals. Blood was collected regularly. Plasma variables were measured and insulin sensitivity was estimated using RQUICKI. Cows were retrospectively categorized according to pre- and postpartum activity into those with the lowest (LOW; 25%), modest (MOD; 50%), and highest (HIGH; 25%) activity. Data were analyzed using a mixed model with repeated measures (fixed effects of activity and time). Nonparametric correlations were analyzed. For activity categorized from d −10 to −4, LOW had higher plasma NEFA on d −15 and 1, relative to HIGH (e.g., 744 vs. 367 µM on d 1; P < 0.01). LOW had decreased RQUICKI values and glucose levels on d −15 and 1, respectively, relative to HIGH (P < 0.01). For activity categorized from d 1 to 7, LOW had higher plasma NEFA on d 1, relative to HIGH (P < 0.01). Plasma BHBA and insulin were not modified by activity. Plasma NEFA levels were negatively correlated with previous 24-h activity on d 1 and 4 (r = −0.33 and −0.26, respectively; P < 0.05). RQUICKI values were positively correlated with previous 24-h activity on d 1 (r = 0.25; P < 0.05). Similar results were observed on d 1 and 4 when evaluating daytime activity. Prepartum activity was negatively correlated with postpartum BCS loss (P < 0.01). Lower prepartum activity of dairy cows is associated with reduced insulin sensitivity, and accelerated NEFA mobilization and BCS loss postpartum. Activity monitoring within large herds has the potential to serve as an inclusive alternative to laboratory NEFA testing.