The theory of system identification was used to determine the time constant tau of a 1 litre flow through differential calorimeter (Setaram GF 108) at a flow rate of 50 ml min(-1). By numerical differentiation the impulse response function g(t), the time derivative of the step response f(t), was calculated. With the aid of the Prony method, the time constant a(2) of the time-discrete system of the decimated dataset was calculated, giving a mean value of 0.7402 +/- 0.0044 (n = 4). This value was converted to the time constant tau of the time-continuous system, giving a value of 33.25 +/- 0.65 min (n = 4). The description of the system agreed with a model for a first order process. For control of the time constant value, the step response f(t) and the impulse response g(t) signal were simulated from the original block diagram u(t) which gave a suitable fit. Via the technique of deconvolution, the datasets of a biological case study with goldfish (Carassius auratus L.) were desmeared to describe the dynamic responses of the biological processes in the calorimetric vessel with a much reduced time constant tau. Finally, the timescale on which the process of metabolic depression takes place in this species during anaerobiosis was estimated to be several minutes.