Ultrasonography was used to determine whether there is embryo reduction in mares iwth unilaterally fixed twins when a major portion of the vascularized area of the wall of one of the embryonic vesicles is in apposition with the wall of the adjacent vesicle, rather than with the endometrium (deprivation hypothesis). In addition, the effect of ovulatory pattern (synchronous and asynchronous) on the incidence of embryo reduction was studied. Twin vesicles were ultrasonically detected on days 11 to 15 (ovulation = day 0) and were examined daily until there was embryo reduction or until day 40. In 31 mares with twin embryonic vesicles, unilateral fixation (71%) was more frequent (P less than 0.05) than was bilateral fixation (29%). In 28 mares with known ovulatory patterns, synchronous ovulations did not affect the type of fixation (9/17 unilateral and 8/17 bilateral); however, for asynchronous ovulators the frequency of unilateral fixation (10/11) was greater (P less than 0.01) than the frequency of bilateral fixation (1/11). The incidence of embryo reduction was greater (P less than 0.01) for unilateral fixation (14/19) than for bilateral fixation (0/9) and was greater (P less than 0.05) for asynchronous ovulators (9/11) than for synchronous ovulators (5/17). In mares with embryo reduction, the reduction was complete before detection of both embryo propers (early reduction) in 10/14 and after detection of both embryo propers (late reduction) in 4/14. For 17 synchronous ovulators, fewer underwent early embryo reduction (0 mares) than late reduction (5 mares) or no reduction (12 mares; 4 unilateral and 8 bilateral), whereas in the 11 asynchronous ovulators, more underwent early reduction (8 mares) than late reduction (1 mare) or no reduction (2 mares; 1 unilateral and 1 bilateral; P less than 0.01). In mares with early embryo reduction, the orientation and spatial relationship of one vesicle relative to the other was not determinable until the embryo proper was detected. In the 2 mares in which the embryo proper of the survivor was detected before embryo reduction was complete, the embryo proper was located opposite to the site of reduction, ie, the vesicle that was to be eliminated impinged on the thin-walled area of the yolk sac wall of the survivor. The position of the embryo proper and its emerging allantoic sac seemed to determine whether a given conceptus survived or underwent late embryo reduction. The embryo proper, the vascularized wall of the yolk sac adjacent to the embryo proper, and the emerging allantoic sac were exposed to the endometrium (uterine lumen) in the surviving vesicles; in the vesicles that underwent reduction, much of the corresponding area of the vesicle wall was covered by the wall of the adjacent survivor. The results supported the deprivation hypothesis.