Endocrine research is frequently dogged by the problem of the signal, in this case hormone level, fluctuating at a frequency at least as fast, if not faster, than the sampling frequency; a situation that would be considered unacceptable in other areas of science. Continuous blood sampling offers a means by which this problem may be circumvented. The technique is illustrated, and its results discussed, together with statistical techniques which have been designed to handle and analyse the repetitive serial data that is obtained by such sampling.