The objective of this study was to determine the effect of quarter-level take-off settings and feeding during milking on milk flow and efficiency within automatic milking systems (AMS). A total of 30 dairy cows (parity 2.9 ± 1.5, 142 ± 25 DIM, milk yield = 34.0 ± 11.7 kg/d and SCC <115 000 cells/mL) were exposed to treatments using a Latin square with 7d periods. Treatments were take-off (TO) at flow rates of 0.48, 0.30 or 0.06 kg/min on a quarter level using the settings of the AMS, combined with or without teaser feed during milking. Cows for which more than 8h elapsed since the last milking were brought to the AMS holding area, resulting in the milking interval 7.5 ± 0.4h. Data on milk yield and flow, and milking times were automatically collected and the last 2d of each treatment period were used for evaluation of treatment effects. An algorithm was developed for defining the decline phase of milk flow and overmilking was defined when milk flow rate falls below a value of 0.06 kg/min. Data were analyzed by ANOVA for a 6 × 6 Latin square with a 2 × 3 factorial arrangement of treatments in a linear mixed-effects model using repeated measures in SAS (v9.4). Preliminary data suggest that feeding did not affect milking time, milk yield, mean flow, the length of the decline phase, or overmilking time (P > 0.05). Increasing TO from 0.06 to 0.48 decreased milking time (5.3 and 4.6 ± 0.28 min respectively; P < 0.001). However, milk yield was not affected by TO (P = 0.30). Mean flow was lower at 0.06 compared with 0.30 and 0.48 (0.72, 0.80 and 0.81 ± 0.004 kg/min respectively; P < 0.001). Decline phase length was longer on 0.06 compared with 0.30 and 0.48 (80.1, 60.9 and 52.2 ± 2.3s respectively; P < 0.001), as were overmilking times (9.3, 1.2 and 0.1 ± 0.05s respectively; P < 0.001). Actual flow at TO was on average 0.0, 0.025 and 0.088 ± 0.093 kg/min for 0.06, 0.30 and 0.48 kg/min respectively (P < 0.001). These data suggest it is possible to reduce milking time by setting a higher take-off level on quarter level in AMS without losing milk yield. Feeding during milking did not affect mean flow or milking efficiency. Overall, milking efficiency of the AMS appears to be linked to take-off level rather than teaser feed.