Ruminal synchronization of energy and protein supply is a strategy suggested by nutrition-ists to increase the efficiency of ruminant metabolic processes such as microbial protein synthesis and to decrease the environmental excretion of nutrients. A so-called synchronization index (SI), formulated as a function of the ratio of nitrogen to organic matter release in the rumen, is widely used to quantify and evaluate ruminal synchronization. A main concern is that SI values obtained from the in situ experimental method are influenced by many random factors and therefore SI value, calculated by the currently-used formula, may not represent actual degree of synchronization. Therefore, in this paper, we comprehensively examine sensitivity of SI to random variations (unwanted measurement errors) due to the experimental conditions variability. Evaluations were performed by means of in situ experiments on 3 ruminally fistulated Holstein lactating dairy cows, where 18 different diets were studied. The diets (g/kg DM) were formulated to a range of CP (150-190), NFC (300-400), NDF (310-410), and ADF (169-246). Based on measured data, a set of synthetic data were also generated to study a range of random variations and their effect on SI values. Our results indicated that SI values are quite sensitive to the random variations (i.e., significantly decrease as random variations increase). This fact revealed that the currently used SI depends not only on degree of synchronization, but also on degree of random variations due to the inevitable experimental conditions variability. Therefore, the original SI is not fully representative of the actual synchronization when having a high degree of variability. To resolve this problem, we propose a novel approach to calculating SI based on a regression analysis on measured data of nitrogen and organic matter release followed by an analytical derivation of SI based on the regression results. The SI values obtained from the new method showed minimal sensitivity to the random variations. The original method yielded a much lower SI than the new method (0.64 vs. 0.92) for our optimum diet (g/kg DM) (NFC = 350, Abbreviations: ADFom, acid detergent fiber expressed exclusive of residual ash; CHO, carbohydrate; CP, crude protein; DM, dry matter; MAE, mean absolute error; MN, microbial nitrogen; N, nitrogen; NEL, net energy for lactation; NDFom, neutral detergent fiber expressed exclusive of residual ash; NFC, non-fiber carbohydrate; OM, organic matter; RMSE, root mean squared error; RUP, ruminally undegradable protein; XPS, xylose protected soybean. NDF = 330, ADF = 183, and CP = 165), showing some degrees of asynchrony for this diet. Therefore, we conclude that the new method provides more reliable SI values for determining and quantifying ruminal synchronization.