Abstract: Background The Pisha Sandstone region on the Loess Plateau experiences intense hillslope erosion. Clarifying how vegetation types respond to distinct rainfall patterns is essential for targeted restoration and watershed management. Methods In the Getuodian watershed, we monitored runoff and sediment from standard runoff plots representing six vegetation/land-use types (Chinese pine forest, sea buckthorn shrubland, natural grassland, planted grassland, cropland, and bare land) during the period 2019-2023. We grouped 42 erosive storms into four rainfall patterns using k-means, based on rainfall amount, I30 (maximum 30-min intensity), duration, and rainfall erosivity. We compared responses across patterns and used structural equation modeling—based on principal components of rainfall and soil variables—to identify drivers. Results 1) A total of 42 erosive rainfall events were recorded and classified into four rainfall patterns. Among them, Class I events—characterized by the largest rainfall amount and I30—produced the highest runoff and sediment yield. 2) Runoff and sediment yield differed significantly among plot types (P < 0.01), with the overall order: bare land > cropland > planted grassland > natural grassland > shrubland > forest. Across rainfall patterns, each vegetation type showed significant differences in runoff and sediment yield (P < 0.05); only the Chinese pine plot remained low even under Class I rainfall. 3) The dominant drivers varied by rainfall pattern: rainfall factors were significant in Classes I, III, and IV (P < 0.05); soil factors were significant in Classes II, III, and IV (P < 0.05); and slope was significant across all patterns (P < 0.05). For runoff, Class I events were dominated by rainfall factors (P < 0.001), whereas Classes II and IV were dominated by soil factors (P < 0.001). For sediment yield, runoff volume was the principal driver under all patterns. Conclusions This study demonstrates how vegetation types and rainfall patterns shape runoff and sediment yield in the Pisha sandstone region and identifies the key driving factors, thereby providing a scientific basis for vegetation restoration and watershed management in this area.