Phosphorus (P) inputs into water reservoirs are the primary cause for accelerated eutrophication affecting water quality. Attempts are underway to regulate inputs originating from concentrated animal feeding operations (CAFOs). The purpose of this research was to relate runoff dissolved (DP) and total P (TP) losses to site-specific characteristics from plots in CAFOs and compare them to their corresponding risk assessment using the Texas Phosphorus Index (PI). Initial studies showed that soil test P (STP) methods used in Texas by inductively coupled plasma were highly reproducible regardless of manure source or application rate. However, NH4OAc-EDTA extraction efficiency was increased with respect to other methods as soil conditions became less acidic, probably due to dissolution of the greater portion of Ca-bound P resulting in STP values that could be three times greater than those of Mehlich III for the same soil. Surface application of dairy manure to high pH soils were positively correlated to STP at various soil-sampling depths down to 15 cm. First order linear relationships between STP values and DP concentrations in runoff were statistically significant for extraction methods and sampling depths but were different among different soils under neutral to calcareous conditions. Attempts to reproduce this relationship on fields that received periodic applications of manure or effluent with various incubation periods failed, although there was a single highly significant relationship between STP and runoff DP for different soils when soil conditions were acid ( pH<6.5) with various sampling depths. Analyses of NH4OAc-EDTA extractable soil elements showed Mg was significantly correlated to DP concentration across various management and soils, indicating that Mg-bound P is a major component controlling P release into runoff. Use of the Texas PI reflected vegetation type closely, with grass-covered sites averaging the lowest risk rating, and having the lowest DP and TP losses, while conversely tilled sites had the highest. However, overall relationship was poor when estimates for erosion rates were used due to experimental design limitations. Use of measured erosion rates for plots and inclusion of extractable Mg improved correlations between PI rating to DP and TP losses, with r2 ranging from 0.60 to 0.87.