Our objectives were to characterize responses in the field to a mix of fibrolytic enzymes (enzyme) using large dairy herds and sufficient study power to evaluate milk production and reproductive responses to enzyme use that began pre-calving. We hypothesized that use of enzyme would increase milk yield, when provided to dairy cattle pre-calving and for approximately 200 d of lactation. There were 7,507 cows in 8 replicates and 16 pens on 3 US dairies. Eight pens were randomly allocated as control pens (received no enzyme), while another 8 received enzyme (750 mL/t of DM feed, with estimated 350,000 and 10,000 units/g of xylanase and cellulase activity, respectively). A mixed model was fitted to milk and bodyweight (BW) data with the fixed effects of treatment, parity, dairy, and herd test and all 2-, 3- and 4-way interactions. The data were repeated measures (herd test 1–5) within experimental units with the random effects of identity, nested within treatment, replicate, and dairy. Milk and energy corrected milk yield were increased by enzyme use by 0.73 and 0.75 kg/d, respectively across a 5-mo period (P < 0.001). Milk fat percent was not increased by enzyme use (P = 0.914), but milk fat yield was increased by 0.04 kg/d, compared with controls (P < 0.001). Milk protein yield increased 0.02 kg/d with treatment (P < 0.001) despite a reduction of 0.02 in milk protein percent (P = 0.008). Log SCC and BW were not influenced by treatment (P = 0.888 and 0.322, respectively). Dry matter intake (DMI) was numerically 0.2 kg/head per d higher for enzyme pens (P = 0.160). Most production responses to enzyme were influenced by dairy. Milk yield in treated cows compared with controls was significantly higher by 3.53 kg/d on Dairy 2 and numerically higher by 0.55 and 0.18 kg/d on Dairies 1 and 3, respectively. Reproduction, health, and risk of removal or death were not influenced by treatment, apart from a reduced time to first breeding (hazard ratio = 1.266; P = 0.001). Production responses to enzyme varied with dairy from substantial to minor increases, but variation among dairies was not evident in differences in DMI nor BW among treated and control pens.