Zero drift may be a cause of imprecision when micromanometer tipped catheters are used for intravascular pressure measurement over long periods of time. Drift of only a few mm Hg may represent a significant error when accurate recording of low vascular pressures is required. To overcome this problem a micromanometer tipped catheter has been modified so that it can be calibrated easily while it is in the circulation. Laboratory testing has demonstrated that when zero drift occurs this intravascular "reference calibration" is a valid linear function of true zero (r = 0.999). As the sensitivity of the catheter is unaffected by zero drift, it is possible to measure pressure accurately by compensating for this zero drift. During dynamic testing of two catheters, there was a mean net drift over 24 h of -0.54 mm Hg. Clinical evaluation of the catheter was undertaken in the human pulmonary circulation in eight patients (two for 48 h, five for 24 h and one for 8 h). In contrast to the laboratory findings, over the first 4 h after catheterisation there was a phase of rapid zero drift: the net drift was -1.9 (SD 3.3) mm Hg with a range of drift of 5.5 (7.4) mm Hg. Subsequently there was gradual drift: the net drift between 4 and 24 h was -0.44 (2.1) mm Hg with a range of drift of 2.8(1.0) mm Hg; and the net drift between 24 and 48 h was 3.7(2.1) mm Hg with a range of drift of 4.1(1.9) mm Hg. During long term intravascular pressure measurement with micromanometer tipped catheters, zero drift may occur unpredictably and should be quantified.