It is suggested that the cumulative view of growth in which some index of biomass is plotted against time tends to obscure temporal variations in the growth process that might provide evidence of how it is controlled. Experiments with the colonial hydroid Campanularia flexuosa show that the action of a growth control mechanism can be demonstrated by considering changes in specific rates of growth determined at frequent intervals in time. However, it is also necessary to disturb the growth process slightly in order to initiate the action of the control mechanism, and having done so, to isolate the effect of the disturbance on growth and thus the action of the control mechanism. This is done by expressing the specific growth rate of organisms whose growth is disturbed as a percentage of that of control organisms of the same age.Experiments in which various toxic agents have been used to disturb growth show that the response of the control mechanism is indiscriminate, and counteracts inhibition irrespective of the agent responsible. The results of further experiments in which colonies were subjected to changes in the level of inhibitor, or in which the inhibitor was removed after the colonies had equilibrated to it, are consistent with the behaviour of a control mechanism operating by negative feedback. Corroborative evidence for this hypothesis comes from work with a simulation model employing feedback. This interpretation of growth control also helps to explain the results of other experiments in which increased resistance to a toxicant resulted from prior exposure to low levels, and others in which ' catch-up' growth and hormesis were observed.