The aim of this study was to develop a methodology to measure sow udder conformation to use in studying the correlation between udder traits and piglet survival, health and performance. The steps in the investigation were (i) to assess the repeatability of measures, (ii) to determine if there was an important difference between the two sides of the udder, (iii) to assess the extent of variation between sows, and finally (iv) to verify if the measures differ in a systematic way over the days shortly before farrowing. A total of 24 sows were scored for six conformation traits of the udder measured twice a day, every day from the sows’ entrance into the farrowing crates until farrowing (1 to 4 days later). The data were recorded from both sides when the sow was lying and when she was standing. The measurements taken were: inter-teat distance within the same row (SAMER; mm between the adjacent teat bases); distance from the base of the teats to the abdominal midline, recorded only in a lying posture (B); distance between the teat base and the adjacent teat on the opposite row, recorded only in a standing posture (OPPR), distance from the base of the teats to the ground (FLOOR); teat length (LEN) measured from the tip to the base, and diameter (DIA) measured at the tip of the teat. Intraclass correlation coefficients (ICC) revealed that most udder conformation traits were highly repeatable (ICC>0.8); only DIA and FLOOR had lower repeatability (ICC=0.7). Measurements did not differ by side. In general, the greatest proportion of variance occurred at the sow level. Traits changed little in the days before farrowing, except for a change 1 day before farrowing in DIA, FLOOR and OPPR. Measures which used anatomical landmarks as the reference point were more reliable than those using the floor of the pen. Udder conformation measures can be used as a reliable phenotype for further study. They can be collected on any day shortly before farrowing, and only from one side and in one posture to save time.