The aims of this study were 1) to determine the effect of operator experience and embryo flushing technique on embryo recovery (ER) in each flushing attempt; 2) to determine the percentage of unwanted pregnancies and endometritis following embryo uterine lavage without administration of PGF2α. Seven non-lactating mares were inseminated with the same semen in four cycles (n=28 cycles). Embryo flushing was performed 7-9 days after ovulation by two operators of different experiences: experienced (>500 flushes) and beginner (<5), using two different flushing techniques in a crossover design. Mares were flushed three times with Ringer's Lactate (RL) “in-and-out”: 1L during the first two attempts and 0.5L during the third attempt. The remaining 0.5L was used to flush the tubing system. The two flushing techniques were 1) uterine massage (UM): continuous ballottement and massage of the uterus per rectum during RL recovery in each attempt, 2) control: the RL was allowed to flow back without touching the uterus during the first two attempts, and during the third attempt (0.5L) the uterus was manipulated as in UM. In both groups, 20 IU of oxytocin were administered at the second attempt and the RL was allowed to remain in the uterus for 3 minutes before recovery. No PGF2α was administered at the end of the flushing procedure. The overall ER was13/28 (46.4%). Ten embryos (10/13,77%) were recovered in the first filter, and 3 in the third filter. The experienced operator obtained a higher (P<0.05) ER (10/14) than the beginner (3/14). The ER of controls (6/14) was similar (P>0.1) to the UM (7/14). More embryos (7/7, P=0.06) were recovered in the first filter in the UM compared with the control group (3/6). Two mares entered prolonged dioestrus following a positive ER. Two mares with a negative ER (13%, 2/15) were pregnant 14 days after ovulation: both were from the control group and flushed by the beginner. In these mares, horn expansion was incomplete after RL infusion in every attempt. The last 8 embryos recovered in the study were transferred using the Wilsher´s technique immediately after recovery in the same donors, but none became pregnant. The overall mean inter-ovulatory interval was 22.5d (16-29d). Six mares (21.4%) developed free-intrauterine fluid and early luteolysis, of which 4 mares had been transferred with their own embryos. In conclusion, the previous operator´s experience had the greatest effect on overall ER. The uterine massage influenced the flushing attempt´s number in which the embryo was recovered. Oxytocin had no effect on ER. The risk of unwanted pregnancies in donors flushed by a beginner with no PGF2α treatment, was considerable.