Mathematical models for predicting nitrogen and phosphorus excretion play a key role in manure application and environment monitoring. An analysis for prediction of fecal nitrogen (FN, g/d) and fecal phosphorus (FP, g/d) excretion for Chinese Holstein lactating dairy cows was conducted using a data set from 15 dairy farms in northern China. The whole independent-variable data set, obtained with questionnaires, consisted of 110 sets of average diet nutrient compositions, including DMI (kg/d), CP content (% DM), OM intake (OMI, kg/d), nitrogen intake (NI, g/d), and phosphorus intake (PI, g/d), and animal characteristics, including average days in milk (DIM, d), average milk yield (MY, kg/d), and average BW (kg). In addition, 110 fecal samples in total were collected to analyze FN and FP excretions, which were considered dependent variables of prediction equations. Correlations between diet and animal variables were examined, and several variable subpools were derived that were used to develop equations to predict FN and FP excretions by stepwise regression analysis. The results showed that among all variables, OMI was the best predictor for FN excretion (root-mean-square prediction error [RMSPE] = 9.58%, = 0.70), followed by NI (RMSPE = 10.19%, = 0.67). However, when both DMI and CP were used as independent variables, the equation showed more accurate prediction for FN excretion (RMSPE = 8.55%, = 0.77) in comparison with univariate prediction equations. Simultaneously, PI was the best predictor of FP excretion (RMSPE = 10.28%, = 0.67). Evaluation results using 3-fold cross validation and comparison with extant equations indicated that the proposed equations were accurate with low prediction errors, which could be recommended for use to estimate FN and FP excretions from Chinese Holstein lactating dairy cows.