The precision and accuracy of National Research Council (NRC 2001) vs. Cornell Pennsylvania Miner (CPM) softwares were evaluated. Sixteen lactating dairy cows (8 in second lactation and 8 in third, with days in milk of 55 ± 31, body live weight = 641 ± 53 Kg, milk yield =31.2 ± 2.5 Kg/day), were assigned to individual pens in a randomized complete block design, with two blocks of two ration treatments (each treatment including 4 cows in second lactation and 4 in the third). The cows were fed rations formulated by NRC 2001 (A) and CPM (B), as totaly mixed rations (TMR) for 90 days. The adaptation consisted of a period of two weeks from the beginning of the experiment. The average milk yields of cows fed with diet A vs. B were 31.11 and 31.9, respectively with no significant difference observed between the two groups for the mentioned trait (p>0.05). There was no significant difference observed for dry matter intake (DMI) in the two ration treatments (p>0.05), the average dry matter intake of diets A vs. B being 22.9 and 22.33 Kg/day, respectively. Both NRC 2001 and CPM underpredicted DMI (-0.63 and -2.43 Kg/day respectively) while the National Research Council model predicting DMI somehow more accurately (in lactating dairy cows) than CPM. Average daily crude protein intakes for diets A vs. B were 3618 and 3438 g/d, respectively (P<0.05). An economical comparison of formulated rations by NRC 2001 vs. CPM (for the costs of feed per unit milk yield) showed a significant difference (1430 vs. 1250 Rails for NRC2001 and CPM, respectively) (p<0.01). CPM predicted a lower protein requirement than NRC 2001 with no negative effect being observed on milk yield or on any other traits. According to the results of the present study, the use of CPM is recommendable in formulating rations for the lactating dairy cows