This experiment investigated the sources of variation in sow udder morphology. A cross-sectional study of 218 sows (109 Large White × Landrace [LWL] and 109 Meidam [Large White × Meishan {MDM}]) of different parities was conducted using a combination of scores and metric measurements. For each teat, 4 measures were taken: the inter-teat distance within the same row (SAMER), the distance from the base of the teat in the upper row to the abdominal midline (AML), the length of the teat from the tip to the base (LEN), and the diameter at the tip of the teat (DIA). Scores were adopted to define teat orientation (0 = teat not orientated perpendicular to the mammary gland and 1 = teat orientated perpendicular to the mammary gland), teat functionality (1 = milk channel not working, including teats that were blind, inverted, or very damaged; 2 = reduced availability of colostrum; and 3 = perfectly functional), and udder development (1 = not developed to 3 = fully developed). A longitudinal study on a subset of sows ( = 70) investigated how udder morphology changed in consecutive parities. Meidam had shorter teats, which were closer to the abdominal midline than LWL (LEN, < 0.001; AML, < 0.001). In both studies, first and second parity sows had smaller teats (LEN, < 0.001; DIA, < 0.001) than older multiparous sows. Teat position had a significant ( < 0.001) effect on SAMER in both breeds, with less distance between middle teat pairs. The distance from the base of the teats in the upper row to the abdominal midline was shorter in the anterior and posterior teats compared with the middle teat pairs. Teat length was greater in the anterior and middle teats than in the posterior ones, whereas DIA was greater in the middle teats. Teat pair position was associated with teat orientation ( < 0.001) and teat functionality ( < 0.001). Parity was associated with udder development ( < 0.001). Breed, parity, and teat pair position were all significant sources of variation in udder morphology in sows.