Catchment liming is one of the main techniques available for treating acidified surface waters. It also provides one of the few tests of models developed to forecast the effects of future changes in stream chemistry. In this study, three experimental catchments were limed between September 1987 and June 1988. Annual biological surveys, covering periods before treatment to the present, were carried out on the manipulated streams and a further 15 reference streams. Empirical models developed from multivariate analysis of pre-liming data were used to forecast the changes in stream biology expected to result from the chemical changes induced by liming. Changes in the occurrence of some invertebrate and plant species indicated clear responses to liming, which matched the model predictions. However, the assemblages as a whole showed marked lags in attaining their final expected status. Trout densities increased after treatment as predicted, but natural variability made the significance of these changes difficult to gauge. The effects of catchment liming on stream trout populations, invertebrates and macroflora, whilst partially consistent with model predictions, remain unclear in many respects. The models may be inadequate, for example, in omitting important variables. Limited replication of sites and runs of pre- and post-treatment data make detection of change against background variability problematical. Thus, discrepancies between model predictions and observed responses may indicate genuine effects of liming, be due to model or sampling errors, or due to disequilibrium in the chemistry and biology of limed streams. © 1992 Academic Press Limited.