The objective of this study was to identify relevant descriptors of ruminal pH post-prandial evolution that can replace the mean pH (considered unsatisfactory). These descriptors are to be used in the attempts to predict ruminal pH from dietary characteristics, in order to quantify the potential of a diet to induce subacute ruminal acidosis from its intrinsic characteristics. A total of 219 pH curves, reported as graphics in 48 published articles describing the post-prandial evolution of ruminal pH (first 8 h), were digitized by image analysis then summarized in 15 pH variables. Relationships among pH variables and the principal components (PCs) of pH variability were analyzed in order to identify possible alternatives to mean pH, as the average value of all pH data the curve is composed of. Two groups of pH variables were identified according to their relationship with the most important principal components. A first group, including mean pH, was closely related to PC 1 , which accounted for 78% of data variability; hence, correlations between variables of this group were generally high. Of these, threshold-related variables were distinct as their withinstudy correlations with mean pH were rather moderate (0.69 on average). This suggests they might carry supplementary information that could explain the variation in ruminal pH induced by within-study factors, e.g. diet characteristics. However, caution should be taken in their use because of their truncation at 0 h and their non-normal distribution. Variables from the second group were independent of the PC 1 , and thus of the first group of variables, whereas they were mostly related to PC 2 and PC 3 . This implies they are complementary to mean pH. Of this second group, the rate of pH decreases or the time period when pH reaches its minimum might be useful to better describe the ruminal status, from the point of view of the risk of subacute ruminal acidosis.