Water-filled tensiometers are widely used to measure the matric potential of sod water. It is often assumed that, because these give a direct reading, they are accurate. With a series of laboratory tests with model laboratory systems of increasing complexity we show that the output of water-filled tensiometers can, particularly in drying sods, be in serious error. Specifically, we demonstrated that water-filled tensiometers can indicate a steady matric potential, typically between -60 and -90 kPa, when the sod is much drier. We demonstrate the use of water-filled tensiometers that can measure matric potentials smaller than -100 kPa in the laboratory and in the field. The physics of the failure of water-filled tensiometers is discussed. When the matric potential was greater than -60 kPa, in laboratory and field tests water-filled and porous matrix sensors were in good agreement. In the field environment the porous matrix sensor was useful because it allowed early detection of the failure of water-filled tensiometers. In dry soils (matric potential < -60 kPa) the porous matrix sensor was more reliable and accurate than the water-filled tensiometer.