Within-farm variation in forage composition can be substantial and potentially costly, and it presents challenges for sampling the forage accurately. We hypothesized that day-to-day variation in forage neutral detergent fiber (FNDF) concentrations and diet variation caused by sampling error would have negative effects on production measures in lactating dairy cows. Twenty-four Holstein cows (73d in milk) were used in 8 replicated 3×3 Latin squares with 21-d periods. Treatments were (1) control (CON), (2) variable (VAR), and (3) overreacting (ORR). On average, over the 21-d period, all 3 treatments were the same [24.7% FNDF and 48.2% forage dry matter (DM) composed of 67% alfalfa silage and 33% grass silage]. The CON treatment was essentially consistent day-to-day in total forage and FNDF concentrations and proportion of alfalfa and grass silages. The VAR treatment changed daily (in a random pattern) in proportion of alfalfa and grass silages fed, which resulted in day-to-day changes in FNDF (range was 21.5 to 28%). The ORR treatment varied in a 5-d cyclic pattern in total forage and FNDF concentrations (26, 24, 28, and 21.5% FNDF). Over the 21d, ORR (25.1kg/d) had higher DM intake compared with CON (24.5kg/d) and VAR (24.3kg/d). Milk production (42.8kg/d), milk fat (3.5%), and milk protein (2.8%) were not affected by treatment; however, a treatment × day interaction was observed for milk production. Lower daily milk yields for VAR and ORR compared with CON were rare; they only followed sustained 4- and 5-d periods of feeding higher FNDF diets compared with CON. In contrast, increased daily milk yields for VAR and ORR versus CON were more frequent and followed sustained diet changes of only 2 or 3d. Lipolytic and lipogenic-related enzyme mRNA abundances in subcutaneous adipose tissue were not affected by treatment. Treatment × day interactions were observed for milk fatty acid markers of cellulolytic bacteria (iso-14:0, iso-15:0, iso-16:0) and lipolysis (18:0) and generally followed the expected response to changes in daily rations. Overall, extreme daily fluctuations in FNDF had no cumulative negative effect on production measures over a 21-d period, and daily responses to transient increases in FNDF were less than expected.