Guidelines to estimate the potential for slope instability and to give new insight into the slope evolution process in regions of deep to moderately deep loess were developed. Laboratory test data indicated a general trend of decreasing cohesion with increasing moisture content for each loess type. Curves for maximum stable slope height to slope angle were produced from the Culmann analysis using average shear strength parameters and unit weight values. Comparison of the theoretical curve for friable loess with empirical data showed good agreement for estimating maximum height at slope angles greater than 50 deg. Stability of slopes with slope angles ranging from 80 deg to 20 deg and corresponding heights calculated from an empirical relationship were assessed using Spencer's method. Slopes with angles between 80 deg and 40 deg had safety factors close to 1.0. For gentler slopes, the safety factors range from about 1.4 to about 2.0 as the slope angle decreases. Empirical data may be a more reliable indicator of potential slope instability than the Culmann analysis over a wide range of slope angles. Analysis of slope failures after one intense rainstorm indicated that friable loess on steep slopes does not become saturated; however, the cohesion is reduced to a level that causes landslides. On gentle slopes, the soil became saturated above the liquid limit, thereby causing mudflows. Although natural events such as this are rare, anthropogenic activity may create slope instability. Analysis of another landslide below a roadway on a friable loess slope indicated a stable to marginally stable condition without surcharge loads, but safety factors dropped below 1.0 when a truckload surcharge was applied. These analyses suggest that natural and cut slopes in friable loess are at safety factors only slightly greater than 1.0. Any construction activity can upset this delicate equilibrium and result in a landslide.