Summary. Blastocysts were flushed out of both uterine horns of gilts on Days 10, 11, 12 or 13. In mated non-pregnant gilts flushing had no effect on progesterone profile or cycle length (20·8 ± 0·4 versus 20·6 ± 0·6 days in the preflush cycle, N = 6, mean ± s.e.m.). Flushing the blastocysts out of the uterine horns on Day 10 resulted in a cycle with a normal progesterone profile and a normal length (21·2 ± 0·4 days, N = 5). Flushing on Days 11, 12 or 13 resulted in a normal cycle or in maintenance of the CL for 3–13 days as indicated by elevated progesterone concentrations and an increased interoestrous interval of, respectively, 22·0 ± 1·2 versus 19·8 ± 0·6 days (Day 11; N = 6), 24·8 ± 1·4 versus 21·0 ± 0·6 days (Day 12; N = 5; P < 0·05) and 26·3 ± 2·3 versus 20·5 ± 0·4 days (Day 13; N = 6; P < 0·05). There was a positive relationship between the change in interoestrous interval and the interval between the first observed standing oestrus and flushing of the blastocysts (rs = 0·350; n = 22; P < 0·1). There was a large variation in the diameter of the blastocysts flushed on the same day. Only in those gilts in which the blastocysts were ≥ 8 mm or filamentous were the CL maintained for 3 or more days. These results indicate that a first signal for maternal recognition of pregnancy is generated on Day 12 and that blastocysts ≥ 8 mm are required for prolongation of CL function for 3 or more days. Since CL function is only extended for a maximum of 13 days (mean 7·4 ± 1·0), a second signal seems necessary to maintain the CL for the whole period of pregnancy. Keywords: CL; blastocysts; pregnancy; progesterone; pig