Heavy metals (HMs) in farmland soils lead to adverse influences on ecosystem and human health. Despite that, data on quantitative risk from different sources are still scarce. In this study, 100 farmland soil samples in Jiuyuan District were collected and analyzed for selected HMs (As, Cd, Co, Cr, Cu, Mn, Ni, Pb, V and Zn) content characteristics and pollution statuses. The positive matrix factorization (PMF) model combined with the Nemerow integrated risk index (NIRI) and human health risk assessment (HHRA) was used to quantitatively identify the primary risk sources. The results indicated that the mean contents or median values (mg/kg) of 10 HMs were all higher than the background values. The contamination factor (CF) and pollution load index (PLI) revealed that the soil was severely polluted. Based on PMF, the main source of HM pollution was anthropogenic activities, accounting for 78.91%. Sewage irrigation represented the biggest input but was not associated with the highest risk. The results of PMF-based NIRI and PMF-based HHRA showed that the chemical fertilizers and pesticides were the largest and priority risk sources with contribution rates of 38.10% to ecological risk and 34.61 and 32.82% to non-carcinogenic and carcinogenic risk, respectively. In addition, non-carcinogenic risk of children was higher than that of adults, while the carcinogenic risk was the opposite. The integrated approaches were beneficial for priority risk quantification from different sources and can provide direct risk information and effective policy recommendations for management and control of key risk sources.