Simulation was used to examine the consequences of ignoring cytoplasmi
c effects on genetic evaluations. Effects on variance component estima
tion, accuracy of selection, genetic trend, and selection of bull-dams
were considered. Comparisons were made for sire and animal models, se
veral levels of cytoplasmic variance, and definition of cytoplasmic ef
fects as either fixed or random. Ten replications with approximately 9
000 cows and 20,000 records were simulated for each model and level of
cytoplasmic variance. Derivative-free REML was used for estimation of
variance components. Ignoring cytoplasmic effects caused overestimati
on of heritability with an animal model. Permanent environmental varia
nce was underestimated with an animal model and overestimated with a s
ire model. Results were used for simulation of large (approximately 20
0,000 cow) data files for analysis of effects on accuracy of selection
. Accuracy of selection increased when cytoplasmic effects were accoun
ted for properly. Improvement of predicted breeding values was greater
for cows than for sires and increased as cytoplasmic variance increas
ed. Genetic trend also increased slightly, primarily as a function of
increased accuracy of bull-dam selection. Little practical difference
was detected when cytoplasmic effects were considered to be fixed rath
er than random.