SUMMARY.1. In regional studies of surface‐water acidification, annual means of chemical variables are often used to describe differences and change. Outputs from hydrochemical models are often in the form of mean values, which are used in biological models, and these, in turn, are usually derived from responses to mean conditions. Thus, biological forecasts are constrained to ignore the possible effects of the short‐term variations in water chemistry which characterize acid streams. This approach requires appraisal.2. Here, regional Welsh data and daily records from a smaller number of streams were used to investigate the estimation of pH parameters. Variations in aluminium concentration in relation to pH were also assessed. Empirical relationships between invertebrate assemblages, fish populations, mean stream chemistry and measures of fluctuation in pH and aluminium concentration were explored.3. In general, pH or Al variability and mean pH or Al in Welsh streams were closely related, so that the biological influences of episodes could not easily be separated from those of chronic conditions. Mean pH and mean aluminium concentration were the most effective pH and aluminium statistics used in multivariate models of trout density, which were not improved by including other pH or aluminium variables. For models of invertebrate assemblages based on mean pH or mean aluminium, the inclusion of variables related to episodicity (e.g. pH minimum, aluminium maximum) gave moderate increases in precision.4. This analysis indicates that it is reasonable to use means of stream chemical variables in biological models of acidification. Consideration of chemical variability could give improvements in some cases, but at the expense of increased model complexity and effort in parameter selection. Nevertheless, we emphasize the need for accurate calibration of both biological and hydrochemical mod