The nonpoint source pollution can be controlled by implementing various best management practices (BMPs) in the watershed. However, before such practices are adopted, their effectiveness at various spatial and temporal scales must be evaluated. Though the effectiveness of individual BMPs has been usually assessed in standard plots, it is necessary to quantify the impact at a wider range (eg. at the watershed scale) to ensure that practices taken will be sufficient to meet EPL (the environmental protection law) targets. In this paper, we compare the characteristics and suitability of different approaches (direct measurement, nutrient budgeting, risk assessment and model simulation) to assess the effectiveness of actions to mitigate sources and transport of nitrogen (N) and phosphorus (P) from agricultural land to water. Nutrient budgets are most commonly used to quantify nutrient management by evaluating inputs and outputs over a defined time period. Automatic calculations under spreadsheets or other user friendly interfaces are used as accounting procedures to obtain the nutrient surplus or deficit values. System boundaries are flexible, resulting in a range of methodologies applicable from plot to national scale including farm gate, soil surface and soil system budgets. The limitation of this method that it currently fail to consider the timing and transport aspects of mitigation and assume a direct causal relationship between potential and actual nutrient loss. Risk assessment procedures quantify the risk of nutrient loss occurring based on the likelihood of nutrient availability and delivery processes coinciding. Small areas may contribute disproportionately large amounts of nutrients and, by ranking vulnerability to loss, mitigation can be targeted to those areas at highest risk. The approach is most applicable to P loss which is explained by annual variations in runoff and the associated erosion of P enriched soils. But assessments demand increased availability of data, and there is a large degree of uncertainty associated with their spatial and temporal dimensions which is difficult to validate adequately. Models are potentially more comprehensive and able to better reflect choice of mitigation at a range of scales. Nevertheless, model calibration and validation is expensive and requires expertise to perform. Large datasets are required for validation which is essential to confirm their ability to quantify actual loss. Monitoring data usually cannot meet the requirement to develop fully distributed models, which decreases the ability to simulate the mechanism of pollutant mitigation in a precise location in catchment with confidence. The disadvantages of individual approaches indicated that these assessment methods should be integrated to maximize their potential usefulness and positive attributes. These will make nutrient inputs utilized most efficiently and site specific actions to reduce nutrient transport and delivery can be targeted at most cost effectively at various scales. Such an integrated decision support system will also more effectively involve the farmers to join the planning of BMPs.