Integrated source apportionment and risk assessment of metals is of great importance for contamination source control and remediation at the regional watershed scale. To identify metal sources and source-specific ecological risks, sediments were collected in the Wen-Rui Tang urban watershed for metal analysis. Risk assessment showed considerable and extremely high risk for Cu and Cd with large spatial variation. Positive matrix factorization model extracted three main sources with reasonable prediction efficacy for metal concentrations. Due to different toxicity coefficients for various metals, the low concentration contribution of source factor 2 (27.2%) contributed 83.7% of total risk, while a high concentration loading for factor 3 (40.8%) only contributed 4.6% to total risk. Predicted source-specific risk was similar to determined risk level for Cr, Ni and Pb; however, Cu and Cd were predicted with decreased risk, while Zn had increased risk. Triangular fuzzy number (TFN) coupled with stochastic simulation showed elevated trend in risk simulation for Cd and Cu when compared with determined risk. The uncertainties for risk evaluation appear to result from spatial variations in metal concentrations. Source apportionment and specific-risk assessment results suggest that different strategies may be required to address mitigation of elevated metal concentrations versus ecological risk.