Various techniques are provided for calibrating and estimating left atrial pressure (LAP) using an implantable medical device, based on impedance, admittance or conductance parameters measured within a patient. In one example, default conversion factors are exploited for converting the measured parameters to estimates of LAP. The default conversion factors are derived from populations of patients. In another example, a correlation between individual conversion factors is exploited to allow for more efficient calibration. In yet another example, differences in thoracic fluid states are exploited during calibration. In still yet another example, a multiple stage calibration procedure is described, wherein both invasive and noninvasive calibration techniques are exploited. In a still further example, a therapy control procedure is provided, which exploits day time and night time impedance / admittance measurements.