CONTEXT Discrepancies at the abutment/crown interface can affect the longevity of zirconia restorations. AIM The aim was to evaluate the marginal and internal discrepancies (MD and ID) of zirconia copings manufactured by two milling systems with different finish lines. MATERIALS AND METHODS Three aluminum-master-dies (h = 5.5 mm; Ψ =7.5 mm; 6), with different finish lines (large chamfer [LC]; tilted chamfer [TC]; rounded shoulder [RS]) were fabricated. Twenty impressions were made from each master die and poured. Sixty zirconia copings were manufactured and divided according to the factors "finish line" and "milling system" (n = 10): CAD LC = Computer-aided design/computer-aided manufacturing (CAD/CAM) + LC; CAD TC = CAD/CAM + TC; CAD RS = CAD/CAM + RS; MAD LC = manually aided design/manually aided manufacturing (MAD/MAM) + LC; MAD TC = MAD/MAM + TC; and MAD RS = MAD/MAM + RS. For MD analysis, each coping was fixed, and the distance between the external edges of the coping and the edge of the cervical preparation was measured (50 points). Using the same copings, the ID of each coping was evaluated, by the replica technique, at 12 points equally distributed among the regions (n = 10): Ray (R), axial (A), and occlusal (Occl). The measurements were performed by optical microscopy (Χ250). The data (μm) were subjected to parametric and non-parametric statistical analyses. RESULTS For the MAD/MAM system, the "finish line" (P = 0.0001) affected significantly the MD median values (μm): LC = 251.80 a , RS = 68.40 a and TC = 8.10 b (Dunn's test). For the CAD/CAM system, the median MD values (μm) were not affected by the factor "finish line" (P = 0.4037): LC = 0.82 a , RS = 0.52 a , and TC = 0.89 a . For the ID, it was observed interaction between the finish line types and the region (P = 0.0001) and between region and milling system (P = 0.0031) (RM-ANOVA). CONCLUSIONS The CAD/CAM system presented lower MD values, regardless the finish line. However, the MAD/MAM system showed ID values smaller than those of CAD/CAM.