Digital levelling ensures high accuracy in height difference determination due to the automated reading of barcode staff images. However, the effectiveness and reliability of the results depend on the spatial orientation of the staff relative to the instrument’s line of sight. This study investigates the influence of barcode staff rotation around the vertical axis on the reading accuracy of the Sokkia SDL30 digital level. An experimental methodology is proposed to quantitatively assess the height measurement errors caused by perspective distortion of the barcode pattern as the rotation angle changes. Measurements were conducted at various distances (2, 4, 6, and 8 m), with each case compared to the reference readings obtained under central alignment. For each data series, regression models were developed to describe the relationship between the measurement error and the rotation angle of the staff. The largest increase in error was observed at intermediate distances, which was confirmed by the slopes of the regression curves and the corresponding coefficients of determination. To generalize the results, a second-degree polynomial model was developed to adequately describe the sensitivity of measurements with respect to staff distance. The obtained results confirm that even a slight rotation of the staff can lead to significant systematic errors in height determination, caused by asymmetry in the image projected onto the CCD matrix. The study establishes the feasibility of introducing a permissible rotation angle threshold to maintain accuracy within the instrument’s technical specifications. The proposed methodology can be applied to improve field observation procedures, calibrate digital levels, and adapt automatic reading algorithms under conditions of complex geometry.