The arterial baroreflex is commonly evaluated by measuring the lengthening and shortening in pulse interval in response to changes in systolic (S) blood pressure (BP) induced by infusion of vasopressor and vasodepressor drugs. This method is simple but has important limitations: only a few observations can be made, reproducibility of the responses is far from optimal, interference by direct drug action on the sinus node and the baroreceptors cannot be excluded and the artificially induced stimuli only poorly mimic the naturally occurring pressor and depressor transients. A new approach was therefore pursued. Blood pressure was recorded intra-arterially for 3 +/- 0.4 h (mean +/- s.d) in 10 unanaesthetized, unrestrained cats and the recording was scanned by a computer to identify the spontaneous sequences of three or more consecutive beats in which SBP progressively rose and pulse interval progressively lengthened (type 1 sequence) or SBP progressively fell and pulse interval progressively shortened (type 2 sequences). Accurate beat-to-beat measurements of SBP and pulse interval were obtained by adopting a very narrow sampling interval of the BP trace, i.e. 1.6 ms real time. For each sequence the regression between the SBP values and the pulse internal values of the following cycle was calculated. In each cat a large number of three-beat sequences were found, the four-, five- and six-beat sequences being, however, progressively less common. All sequences had a high correlation coefficient (r > 0.9), type 1 having a greater slope than type 2 (14.1 +/- 2.5 versus 10.3 +/- 7.6 ms/mmHg, P < 0.01).(ABSTRACT TRUNCATED AT 250 WORDS)