Rate of genetic improvement for milk yield has been increasing in recent years. Cows born in 1986 were about 135 kg superior in breeding value for milk yield to those born in 1985. Over 2.2 million cows contribute new data to genetic evaluations for production traits annually. These evaluations are computed with an animal model that provides best linear unbiased predictions of transmitting abilities for milk, fat, and protein yields and fat and protein percentages. The model includes effects of management group, permanent environment, herd-sire interaction, and animal genetic merit. Unknown-parent groups represent the genetic merit of base populations defined by birth year and sex. Type appraisal data are collected by breed associations and are evaluated with a sire model. Holstein cow evaluations are computed using scores from all appraisals and a multitrait model; evaluations for other breeds are computed using all appraisal scores, a repeatability model, and a single-trait system. Dystocia data are collected by individual AI organizations and dairy records processing centers; they are analyzed by a categorical-trait sire model at Iowa State University with support from the National Association of Animal Breeders. The AI organizations have been extremely important in increasing rate of genetic progress by increasing numbers of young bulls sampled, increasing selection intensity of bull dams through multiple ovulation and embryo transfer, and shortening generation interval through the use of younger cows and some virgin heifers as bull dams. (ABSTRACT TRUNCATED AT 250 WORDS)