Soil matrix suction is assuming an increasing importance in geotechnical engineering. It is now recognised that soil suction is not confined to arid or semi-arid regions of the world. It is frequently encountered in a great many geotechnical problems in temperate climates. For example, all fills and compacted soils have an inherent suction when they are formed; many slopes, both natural and man-made, have in-situ suctions; and all samples that are removed from the ground during a site investigation possess a soil suction in their unconfined state. Measurements of soil matrix suction can be used to: (a) identify disturbance and/or desiccation in samples removed from the ground; (b) estimate the average in-situ effective stress once the effect of disturbance and/or desiccation has been quantified; (c) apply more rigorous controls to the compaction of fill materials where swelling is likely to cause problems; and (d) assess the stability in natural and engineered slopes. This paper describes and compares three methods of matrix suction measurement: the filter paper method, the Imperial College suction probe, and a new in-situ tensiometer that is well suited to long-term observations of in-situ suctions up to about 90 kPa. The principles of their operation, the procedures by which they can be routinely used, and advice on how the results can be interpreted, are presented.