The Loess Plateau (LP) of China is a globally recognized ecologically fragile region, where concentrated precipitation and frequent heavy rainfall events contribute significantly to severe soil erosion. Although precipitation changes on the LP have been widely studied, their underlying causes remain unclear. This study investigated the spatiotemporal variations of precipitation and extreme precipitation across the LP from 1980 to 2018, explored their links to internal climate variability, and quantified the effects of external forcings using CMIP6 Detection and Attribution Model Intercomparison Project experiments and the optimal fingerprinting method. Results showed that annual mean precipitation exhibited a slight upward trend, with a more pronounced increase after 2000 (3.9 mm year-1), particularly in the northern LP. Four of six extreme precipitation indices (R95p, R99p, R10, R20) also showed increasing trends. Wavelet coherence analysis revealed that these changes were strongly associated with internal variability, especially El Nino-Southern Oscillation (ENSO), while Atlantic Multidecadal Oscillation (AMO) and Arctic Oscillation (AO) influenced extreme precipitation during specific periods. Attribution analysis revealed that natural forcing played a positive role in the observed increase in precipitation. Among anthropogenic forcings, greenhouse gas emissions were associated with a positive contribution to precipitation, whereas aerosols exerted predominantly negative influences across most regions. However, the overall anthropogenic signal was not clearly detected, likely due to the offsetting effects of different drivers and the masking influence of strong internal climate variability. This study enhanced the understanding of precipitation dynamics over the LP and provides a methodological framework for regional-scale attribution analysis.