This study evaluated the prediction accuracy of grass dry-matter intake (GDMI) and milk yield predicted by the GrazeIn model using a large database representing 8787 per cow GDMI measurements. In this study, the animal input variables (age, parity, week of lactation, potential peak milk yield, milk fat content, milk protein content, bodyweight, body condition score (BCS), week of conception, BCS at calving and calf birth weight) were investigated. The mean actual GDMI of the database was 159kg DM per cow d(-1) and GrazeIn predicted a mean GDMI for the database of 155kg DM per cow d(-1). The mean bias was -04kg DM per cow d(-1). GrazeIn predicted GDMI for the total database with an RPE of 155% at cow level. The mean actual daily milk yield of the database was 213kg per cow d(-1) and GrazeIn predicted a daily milk yield for the database of 222kg per cow d(-1). The mean bias was +09kg per cow d(-1). GrazeIn predicted milk yield for the total database with an RPE of 167% at cow level. From the evaluation, GrazeIn predicted milk yield of all cows in late lactation with a larger level of error than in early and mid-lactation. This error appears to be due to the persistency of the lactation curve used by the model, which results in a higher predicted milk yield in late lactation compared with the actual milk yield.