The importance of effects of cytoplasmic inheritance on milk production traits of dairy cattle was examined. Three approaches were taken;Effects of cows' maternal lineages on production were estimated by using animal models. Maternal lineages were established on the basis of cows' earliest recorded female ancestors. Data were 6054 records from 2264 cows from North Carolina and Iowa research herds. First parity records from North Carolina, all records from North Carolina, and all records pooled from Iowa and North Carolina were analyzed by separate procedures. Both yield and milk composition traits were studied. Significant associations with maternal lineages were detected for milk fat percentage and milk energy concentration. Variance components for maternal lineages were also estimated. Maternal lineages accounted for 2.9% of phenotypic variance in fat percentage found in the pooled records. Otherwise, 1% or less of the variance in any trait was associated with maternal lineages;The sequences of the displacement-loop and ribosomal RNA genes of mitochondrial DNA were determined for cows from North Carolina and Iowa herds. Effects of polymorphism in these regions of DNA on production traits were estimated by using animal models that included regression coefficients for sites of polymorphism. Significant associations between phenotype and mitochondrial DNA polymorphism were detected at several sites for most traits. Without exception, when a significant effect existed, the wild-type sequence was favorable to the mutant (less common) type;Simulation was used to determine impacts of ignoring cytoplasmic effects on estimation of variance components, accuracy of genetic evaluations, genetic response, and selection of bull dams. Three levels of maternal lineage variance were considered: 2.5, 5, and 10% of total variance. Failure to account for cytoplasmic effects caused reduced selection accuracy and genetic response. Sizes of progeny test programs can be reduced without sacrificing response if cytoplasmic effects are correctly considered. Impacts increase linearly as size of cytoplasmic effects increase. Including maternal lineage effects in a genetic evaluation model when true effects are zero causes reduced accuracy and genetic response. No practical differences were observed whether maternal lineage effects were considered fixed versus random.