This paper is concerned with the response of fluid temperature within a heated pipe to imposed excursions of flow rate. Experiments are reported in which measurements of wall temperature and local fluid temperature were made with fully developed turbulent flow of water in a uniformly heated tube during and after ramp-up excursions of flow rate between steady initial and final values. The fluid temperature measurements were made using a traversable temperature probe incorporating a thermocouple capable of responding to turbulent fluctuations of temperature. Local values of mean temperature and RMS temperature fluctuation were obtained by ensemble averaging the results from many tests in which the same flow excursion was applied in a very repeatable manner with fixed values of inlet fluid temperature and heat flux. Further measurements were made under conditions of steady flow rate at a number of values over the range covered in the transient flow experiments. The results obtained in the experiments with transient flow show that there is a significant delay in the variation of ensemble-averaged wall temperature and striking perturbations in the variations of RMS fluctuation of wall temperature and local fluid temperature. These stem from the delayed response of turbulence to the imposed excursions of flow rate. They provide independent confirmation of ideas concerning the modelling of time scales for the production and diffusion of turbulence in pipe flow which were developed by the present authors in the course of earlier work. Ensemble-averaged local fluid temperature also varies in an unusual manner. Instead of falling monotonically with increase of flow rate, as might be expected, it starts to rise at some stage, reaches a peak value and then falls again. The release of heat stored in the pipe wall contributes to this behaviour. Computational simulations of the present experiments were performed using a spatially fully developed formulation of the equations for unsteady turbulent flow and heat transfer in a boundary layer utilising turbulence models of low Reynolds number, k-epsilon type. Comparisons between predicted and measured variations of temperature are presented in the paper. These show that the predictions differ significantly from model to model and that detailed agreement with experiment is not obtained using any of the models. However, certain interesting features of the observed temperature variations, such as a delay in the response of outer wall temperature and perturbations in local fluid temperature, are present in the computed results. (C) 1999 Elsevier Science Inc. All rights reserved.