Soil acidity is well known to affect the species composition of natural vegetation. The physiological adaptations of plants to soil acidity and related toxicity effects and nutrient deficiencies are, however, complex, manifold and hard to measure. Therefore, generally applicable quantifications of mechanistic plant responses to soil acidity are still not available. An alternative is the semi-quantitative and integrated response variable ‘indicator value for soil acidity’ (Rm). Although relationships between measured soil pH and Rm from various studies are usually strong, they often show systematic bias and still contain high residual variances. On the basis of a well-documented national dataset consisting of 91 vegetation plots and a dataset with detailed, within-plot, pH measurements taken at three periods during the growing season, it is shown that strong spatiotemporal variation of soil pH can be a critical source of systematic errors and statistical noise. The larger part of variation, however, could be explained by the moisture status of plots. For instance, Spearman's rho decreased from 93% for dry plots and 87% for moist plots to 59% for wet plots. The loss of relation between soil pH and Rm in the moderately acid to alkaline range at increasingly wetter plots is probably due to the establishment of aerenchyma-containing species, which are able to control their rhizosphere acidity. Adaptation to one site factor (oxygen deficit) apparently may induce indifference for other environmental factors (Fe2+, soil pH). For predictions of vegetation response to soil acidity, it is thus important to take the wetness of plots into account. Copyright © 2012 John Wiley & Sons, Ltd.