Colloidal particles in runoff may have an important role in P transfer from soils to waterbodies, but remain poorly understood. We investigated colloidal molybdate-reactive phosphorus (MRP) in surface runoff and water extracts of calcareous arable soils from the semiarid western United States. Colloidal MRP was determined by ultrafiltration and operationally defined as MRP associated with particles between 1µm and 1 nm diameter, although a smaller pore-size filter (0.3 nm) was used to define the lower size limit of colloids in water extracts. In surface runoff from three calcareous soils generated by simulated sprinkler irrigation, colloidal MRP concentrations ranged between 0.16 and 3.07 1LM, constituting between 11 and 56% of the MRP in the <1-gm fraction. Concentrations were strongly correlated with agronomic and environmental soil-test P concentrations for individual soils. Water extracts of a range of similar soils contained two size fractions of colloidal MRP: a larger fraction (L0-0.21Lm) probably associated with fine clays, and a smaller fraction (3-0.3 nm) probably associated with Ca-phosphate minerals. Colloidal MRP was solubilized in the acidic medium of the colorimetric detection procedure, suggesting that a fraction of the filterable MRP in runoff from calcareous soils may not be as readily bioavailable as free phosphate in waterbodies. Our results suggest that colloidal MRP is an important but poorly understood component of P transfer in runoff from calcareous western U.S. soils and should be given greater consideration in mechanistic studies of the P transfer process. P HOSPHORUS TRANSFER in runoff from agricultural soils to watercourses can contribute to blooms of toxin producing cyanobacteria (blue-green algae) and other water quality problems associated with eutrophication (Foy and Withers, 1995;Leinweber et al., 2002). As a result, a considerable research effort has aimed to quantify P transfer from agricultural land, and data now exist for a range of soils and agronomic management practices (for a recent review see Haygarth and Jarvis [1999]). Meaningful interpretation of such data requires a thorough understanding of P speciation and the analytical procedures involved in its determination.Conventionally, it is considered desirable to know the concentration of free phosphate in runoff, because this is the form most readily available to algae in waterbodies (Reynolds, 1984). This is estimated by crude fractionation based on membrane filtration (typically 0.45-or 0.2-1.Lm pore size) and reaction with molybdate (Murphy and Riley, 1962), but rarely provides an accurate value of the true dissolved phosphate concentration in envi-