Ergonomic risk assessment methods are widely utilized in the construction sector to mitigate work-related musculoskeletal disorders (WMSDs). However, current methods, such as the Rapid Entire Body Assessment (REBA) and the Rapid Upper Limb Assessment (RULA), often encounter issues related to overestimation and abrupt risk fluctuation, primarily due to sharp boundaries defined by ambiguous terms. Specifically, a 0° threshold is frequently used as a strict boundary for adjustment factors, disregarding minor joint movements and muscle activation. This study aims to quantify the arm-related risk adjustment factors in REBA and RULA by analyzing angle tolerances through muscle activation data. Angle tolerances were derived using surface electromyography (sEMG) data from specific muscles during three motions: arm abduction, wrist deviation, and wrist twisting. A total of 25 participants volunteered for this study. The results indicate that 1) Angle tolerances, defined by the endurance limit of 10% maximal voluntary isometric contraction (%MVC), are different between subjects; 2) minor movements in the shoulder and wrist joints at the onset of motion demonstrate that a 0° threshold is unsuitable as an adjustment boundary; 3) Maximum joint angles of wrist deviation and wrist twisting vary between subjects. Thus, normalization processes are conducted to provide a general result of angle tolerances. Finally, these tolerances were subsequently used to refine posture categorization within the risk rating adjustment process, resulting in a modified REBA and RULA framework incorporating posture sway for more accurate risk estimation.