What are EPDs and How Are They Calculated


Michigan State University Extension
Joel Cowley, Extension Beef Specialist
Department of Animal Science
July 22, 1998

What are EPDs?

Expected Progeny Differences (EPDs) are the most current and accurate means to select cattle for the traits for which they are calculated. It has been suggested that selection based upon EPDs is five to nine times more accurate than selection based upon performance indexes and ratios.

EPDs are estimates of how a bull or cow’s future progeny will perform, on average, for a given trait. The words ‘on average’ are italicized, as this is a very important concept to keep in mind. A parent contributes only a sample half of its genes to each offspring. That sample, being random, might contain a large number of genes that have a positive effect on the trait in question (a good sample) or it could contain many genes that have a negative effect (a poor sample). This can be likened to a deck of cards in a poker game. Some hands are winners and some are losers, but they all come from the same deck. Therefore, EPDs do not predict the absolute performance of an animal’s offspring, but the average performance.

As an example, assume that Sire A has a Birth Weight EPD of +5lb. and the Birth Weight EPD of Sire B is —1 lb. We would expect the offspring of Sire A to average six pounds heavier at birth than the offspring of Sire B when the two sires were bred to a large number of comparable cows (i.e. 94 vs. 88 pounds). Not every calf sired by A will be heavier than every calf sired by B, but Sire A’s calves will be heavier on average.


How are EPDs calculated?

EPDs are usually calculated twice a year when a breed association gathers performance and pedigree information for their breed. The information is sent to an educational institution where a National Cattle Evaluation (NCE) is performed. Most NCEs currently utilize a multiple trait animal model to statistically analyze the data and generate EPDs. An animal model produces an EPD for every animal in the analysis, parent or non-parent, male or female. Animal models take into account all genetic relationships within a data set so that an animal’s own performance is combined and properly weighted with the performance of relatives (progeny, parents, grandparents, full and half-siblings, etc.) in order to generate an EPD. Multiple trait animal models take into account the genetic relationships that may exist between two or more traits and utilize these relationships as another source of information on a trait. As an example, weaning weight information can be used to help calculate Yearling Weight EPDs, as some of the same genes that affect weaning weight also have an influence on yearling weight. This can help to compensate for biases that might occur as a result of sequential culling (culling a sire’s offspring at weaning so that they have no yearling weight data) or selective reporting of yearling data.

The animal model approach also adjusts for the merit of mates. If a sire was mated to only the best cows, his EPD is adjusted to account for this so that he does not receive all of the credit for a superior set of calves.

The genetic change within a breed is also accounted for in an NCE. Therefore, comparisons may be made across generations of cattle. Based upon the available information, young bulls with no progeny can be directly compared with older bulls with a large number of progeny, meaning that more conservative estimates of an animal’s genetic worth are made when information is limited.

The animal model also separates out the maternal component of a trait. As an example, weaning weight is partitioned into the genetics for growth to weaning age (weaning weight direct) and the influence of mild (the maternal component of weaning weight).

It is important to remember that the EPDs generated by any NCE are only directly comparable with EPDs from the same evaluation (within the same breed).

For what traits are EPDs calculated?
EPDs are calculated for a number of, but not all, economically important traits. Most breeds report EPDs for Birth, Weaning and Yearling Weight as well as Milk. The following is a list of traits for which EPDs are calculated.

Birth Weight- Birth Weight EPDs are expressed in pounds and predict the average difference that can be expected in an animal’s offspring when compared with another animal in the same genetic evaluation. Birth weight EPDs are primarily used as an indicator of calving ease, with the age and size of the females to be bred usually dictating how much birth weight can be tolerated.

Weaning Weight- Weaning Weight EPDs are expressed in pounds and predict the average differences in weight that can be expected between the progeny of animals in the same genetic evaluation at 205 days of age. Weaning Weight EPDs do not account for differences in weaning weight that are due to milk.

Yearling Weight- Like Birth and Weaning Weight EPDs, Yearling Weight EPDs are expressed in pounds and predict the average differences that can be expected between the progeny of animals at one year of age.

Milk- Milk EPDs are expressed as pounds of calf weaned by a bull’s daughters. They reflect the average differences in weaning weight that can be expected in grandprogeny due to the milking ability of a bull’s daughters. Available feed resources will dictate the extent to which milking ability should be selected.

Total Maternal (Maternal Weaning Weight)- Like Milk EPD, Total Maternal EPDs are also measured in pounds of calf weaned by an animal’s daughters. They account for average differences that can be expected from both weaning weight direct as well as from milk, and measure a sire’s ability to transmit milk production and growth rate through his daughters. They are calculated by adding an animal’s Milk EPD to one-half of its Weaning Weight EPD.

Yearling Hip Height- Reported in inches. Predict the average difference in progeny hip height that can be expected at one year of age. (Angus).

Calving Ease Direct- Predict the average difference in ease with which a sire’s calves will be born when bred to first-calf heifers. Expressed as percentage of unassisted births with a higher value indicating greater calving ease (Gelbvieh, Simmental, Tarentaise).

Calving Ease Maternal- Predict the average ease with which a sire’s daughters will calve as first-calf heifers when compared to the daughters of another sire in the same evaluation. Expressed as percentage of unassisted births (Gelbvieh, Simmenal, Tarentaise).

Scrotal Circumference- Estimate the average differences that can be expected in scrotal circumference in male progeny. Expressed in centimeters. Of interest as larger scrotal circumference is favorably associated with fertility and age at puberty in a sire’s daughters (Limousin, Angus, Hereford).

Gestation Length- Predict average differences in gestation length. Expressed in days. Shorter gestation lengths are associated with less dystocia and longer post-partum intervals (Limousin and Gelbvieh).

Stayability- Expressed as the probability that an animal’s daughters will remain in production to at lease six years of age when compared to the daughters of another animal. A measure of sustained fertility that probably reflects traits such as fleshing ability and structural soundness. Expressed as deviations from a 50% probability, a higher value indicates increased stayability (Red Angus, Limousin).

Mature Daughter Height and Weight- Predict the average differences that can be expected in mature daughter size in inches and pounds, respectively. These EPDs can be used to match mature cow size to forage resources (Angus).

Carcass Weight- Estimate average differences in carcass weight. Expressed in pounds at a given age endpoint (Angus and Simmental).

Marbling- Predict the average difference in USDA Quality Grade in an animal’s progeny when compared to the progeny of another animal at a given age endpoint. Expressed in numerical marbling score where one point equals one USDA marbling score (Angus and Simmental).

Ribeye Area- Predict the average difference in ribeye area in an animal’s progeny when compared to the progeny of another animal at a given age endpoint. Expressed in square inches (Angus).

Fat Thickness- Estimate the average differences that are expected in fat thickness at the 12th and 13th rib between progeny of different animals. Expressed in inches at a given age endpoint (Angus).

Percent Retail Cuts- Predict the average differences in cutability that can be expected between the progeny of animals at a given age endpoint. Expressed as percent (Simmental).

Docility- Predict the percentage of an animal’s offspring that are expected to score favorably (1 or 2) on a five-point scoring system when compared to the offspring of another animal. Expressed as a percentage with higher values being favorable (Limousin).

How do I know if EPDs are high or low?
It is easy to determine which animal has the highest or lowest EPD, but what about the magnitude of the estimate relative to the breed? The genetic composition of the herd as well as the environmental conditions and marketing strategy, will determine the level to which traits should be selected. Breed averages can be used as a benchmark to determine where an animal ranks within the breed as well as whether or not an animal’s offspring will be suitable for a given set of environmental conditions. Average EPDs will vary between breeds due to differences in Reference Year (the year that all EPDs were arbitrarily set to zero) and Genetic Trend (what has happened in a trait since the Reference Year). Breed averages can be calculated for the entire breed or for a subset of animals. Table 1 lists breed average EPDs for animals born in 1997.   

Table 1. Average EPDs for Animals Born in 1997































Maine Anjou





Red Angus




















*Adjusted to reflect the Fall 1998 change in base 


Another method for comparing breeds based on genetic merit lies within the databases of several breed associations. The policy of many associations allows for the upgrading of percentage cattle to purebred status. As a result, these breed associations have performance in formation of several breeds within their database, information where animals of different breeds are competing within the same contemporary groups. The American Simmental Association implemented a Multiple-Breed Evaluation (MBE) system in their Fall 1997 genetic evaluation. All animals in their database were evaluated in the same statistical model. This allows the resulting EPDs to be directly compared regardless of breed. Such an evaluation would prove beneficial in defining genetic differences for commercial cattlemen. 

What does the Accuracy Value mean?

All EPDs are associated with an accuracy value that represents the reliability of the estimate. Accuracies are computed based upon the quantity and quality of information that went into the estimate and fall in a range between 0 and 1. An accuracy close to 1 indicates that an EPD merits a high level of confidence as it is the result of a great deal of information and is likely an accurate representation of the animal’s true, unknown average genetic value. Values close to 0 indicate a low level of reliability (or a high level of risk) associated with the estimate.

In addition to numeric accuracies, Interim (I) or Pedigree Estimate (P or PE) accuracies may be attached to an EPD. Interim EPDs are calculated when an animal’s own performance was not available or was edited from the most recent NCE. They are computed by averaging the parent’s EPDs and factoring in the animal’s own performance and contemporary group information. Pedigree Estimate EPDs are merely the average of the parental EPDs are do not include the animal’s own performance. Depending upon the trait, P(E) or I accuracies will generally correspond to numeric accuracies of from .10 to .30.

Higher accuracy values indicated that an EPD is less likely to change in subsequent evaluations. Sire summaries contain a table of Possible Change Values to allow you to determine the risk associated with different levels of accuracies for different traits.

A common misconception associated with accuracy is that a bull’s progeny will become more uniform as his accuracies increase. The bull is not changing as his accuracy values increase. He is still the same bull, throwing a random sample of the same genes and producing the same amount of variation in his offspring. The higher accuracy value simply indicates that we are more confident as to what those genes are and, therefore can more accurately predict what the average of his progeny will be when compared with the progeny of another sire.

Is there a way to compare EPDs across breeds?
Scientists at the U.S. Meat Animal Research Center (MARC) in Clay Center, NE have developed an adjustment procedure to convert EPDs from one breed(s) to a comparable scale for another breed. Actual performance differences between the breeds were determined through the comprehensive Germplasm Evaluation Project (GPE), initiated in 1969. Breeds in the GPE are evaluated by artificially inseminating Hereford and Angus cows to a sample of bulls from each breed. The resulting offspring are evaluated for several economically important traits.

By using the actual breed differences (in pounds) from the GPE and adjusting all data to a common base year, adjustment factors are developed to compare the EPDs of cattle of different breeds. Although Across-Breed EPDs appear to merit confidence for growth traits, Milk EPDs may be another matter due to the difficulty in measuring the trait. Other concerns with Across-Breed EPDs appear to merit confidence for growth traits, Milk EPDs may be another matter due to the difficulty in measuring the trait. Other concerns with Across-Breed EPDs are that they do not account for differences that may occur as a result of hybrid vigor and that it is unknown whether or not the same breed differences seen at MARC will hold true across other environments. The estimates listed in Table 2 are not EPDs. They are the most recent MARC adjustment factors that can be added to the EPDs of animals of different breeds in order to adjust them to an Angus equivalent (1996 base). They account for the breed differences observed at MARC as well as the differences in breed average EPDs. As an example, if we wanted to compare the average Angus, Charolais and Hereford animals (from Table 1) with regard to Birth Weight EPD, we could use the factors from Table 2 to adjust all three to an Angus equivalent. Birth Weight EPDs would be 3.1 (3.1 + 0.0), 12.9 (1.7 + 11.2) and 8.1 (4.1 + 4.0) pounds for the average Angus, Charolais and Hereford, respectively. Keep in mind that EPDs are not calculated for every economically important trait and there are reasons for choosing a breed other than just for birth weight, weaning weight, yearling weight, and milk.


Table 2. Across-Breed EPD adjustment factors to adjust breeds to a 1996 Angus Equivalent































Maine Anjou





















Herdbook How To's

Let us know how we are doing