Experimental methods to supply nutrients to culture solutions in order quantitatively to control plant nutrition are compared. In experiments with tomato and birch plants, for which the data are available in databases (Ingestad et al., 1994a, b), the nutrients were supplied at constant relative addition rates (RA over sufficiently long periods of time to achieve acclimated plants and reliable measurements of plant responses. The plants were maintained under steady-state conditions, i.e. the internal nutrient concentrations (c1) remained constant, as a result of a numerical equality between the relative uptake rate (RU) and the relative growth rate (RG). These results are compared to experiments with pea plants (Macduff et al., 1993). In one series (a), RA was applied, but without strict control of internal steady-state, and in the other series (b), the external concentration (ce) was maintained constant. With limiting nitrogen, in both series, there was a substantial deviation from equality between RU and RG. In (a), cI changed during the experimental period and the purpose of the RA approach was lost. In (b), a constant ce had little effect on nitrogen uptake and plant growth. At the three highest concentrations, steady-states were obtained at non-limiting uptake rates. At the lowest concentration, the uptake rate of nitrogen was about the same, but there was a decrease of Ra, which apparently was not caused by reduced uptake. Clear-cut relationships can not therefore be established between treatment variables and plant responses and the conclusions reached by Macduff et al. (1993) have little support in their experimental results. This indicates an urgent need to update both theories and experimental methods together: in particular, it is important to identify the system under investigation and to distinguish between control of the medium and control of the plant.