Breeding schemes for ivool sheep based on natural service (N), MOET (multiple ovulation and embryo transfer) in adults (A) and MOET in juveniles (J) are compared using Monte Carlo simulation allowing overlapping generations and selection across age groups. In all schemes parents can be a maximum of 6 years old when their offspring are born, and can be a minimum of 2 years in N or A schemes or 1 year in f schemes. Embryo transfer success rates follow realistic distributions (only 70% of flushes resulting in live lambs) with a mean of 2·13 or 345 6-month-old lambs per donor per flush for current or ‘future’ situations. Under natural mating an average of 1·09 6-month-old lambs per ewe mated is assumed. The trait under selection has a heritability of 0·4 and is measurable on both sexes by 18 months of age, which is before selection in N and A schemes but after the initial opportunity for selection in f schemes where selection is based on a pedigree index. Two culling policies for males are considered in J schemes: all sires culled if not selected (J), or all sires available for selection after their record is collected even if they were not selected on their pedigree index in the previous year (f o ). Comparison of schemes is made on the basis of response depressed by the effect of inbreeding and cumulated and discounted over a medium time horizon. For a given population size (constant expected number of lambs per year and therefore numbers of dams dependent on fertility rates) different numbers of sires were utilized. The optimum number of sires was determined as the number which maximized the response criterion. N, A, J and J° schemes have different optimum numbers of sires (5, 8,30 and 20 respectively) which occur at different rates of inbreeding and which are approximately independent of population size. Advantages in the response criterion of A, J and J° over N schemes are 23%, 33% and 47% respectively for ‘future’ embryo transfer success rates and mass selection. Increasing the number of lambs born per donor ewe from 2·13 to 13·8 (by increasing embryos per flush and number of flushes) increases the response criterion by up to 25% in A and up to 11% in J schemes (assuming donor ewes are mated to a different sire for each flush) with 345 lambs born per year. Use of best linear unbiased prediction (BLUP) estimated breeding values increases response by 8%, 3% and 3% in N, A and J schemes respectively. The advantages off over A and N schemes, of increased lambs per donor and of BLUP over mass selection all increase as population size increases. Insemination of the poorer ewes with semen which gave only female offspring gave 5% increase in the N scheme and no benefit in A or f schemes.