Many phosphorus risk indicators have been developed to assess site vulnerability to water contamination. These indicators employ a semi-quantitative approach to assess the impact of agricultural management activities on water quality and can be used by landowners and public agencies to target mitigation efforts. Most of the indicators are designed to predict phosphorus loss from land where rainfall-induced erosion of particulate P over sloping landscapes is the main process of P loss from agricultural land. However, there is relatively little information regarding snowmelt-driven losses of soluble P over nearly level landscapes, the main process for P loss on the Canadian Prairies. Furthermore, very few indicators have been validated with water quality measurements and none have been validated in the Canadian Prairies. Therefore, the main objective of this project was to validate several phosphorus risk indicators, including the Minnesota P Index and Canada's National Indicator of Risk of Water Contamination by Phosphorus (IROWC-P), using water quality data from fourteen regional watersheds in Manitoba, Canada. Overall, neither the P index nor the IROWC-P was significantly correlated with mean concentrations of total P in water (r=0.43 and 0.31, respectively) or P export per hectare (r=0.08 and -0.04, respectively). Part of the reason for these poor correlations appears to be the large influence of erosion risk on these indicators, relative to the small influence of erosion on P loss in the watersheds. For example, erosion played a highly significant and dominant role in determining the variability of the P index and IROWC-P values for the watersheds (r=0.65** and 0.66**, respectively) but no significant role in determining mean concentrations of total P in water (r=0.02 for both indicators) or P export per hectare (r=-0.16 for both indicators). However, several site characteristics correlated well with water quality parameters. For example, soil test P values were very highly correlated with mean concentrations of total P in water (r=0.80***), but not with P export per hectare (r=-0.03). Measures of agricultural production intensity, such as livestock density and proportion of land in annual crops were also strongly correlated with P concentrations in water (r= 0.63** and 0.76***, respectively), but not with P export. Similarly, measures of the land's natural characteristics, such as its inherent capability ranking for agricultural production and Manitoba's proposed water quality management zones, were highly correlated with P concentrations in water (r=-0.70** and -0.78***, respectively), but not with P export. In conclusion, existing P risk indicators appear to be focused towards the process of P loss by rainfall induced erosion of particulate P. More research is needed to provide a reasonable estimate of a site's potential for phosphorus loss to surface water, under the soil, landscape and climatic conditions of the Eastern Prairies, where snowmelt-induced runoff of dissolved P is the predominant form of P loss.