An evaluation of the linear description of type in 494 horses of the Old Kladrub breed for 32 traits was used to analyse the effect of variety, stud, sex, year of birth and age at description. A linear model with fixed effects was used. The highest coefficient of variation was calculated for the forelimb side-view (40.14%), chest girth (36.25%) and height at withers (30.97%). 28 out of 32 traits were in the range from 7 to 9 utilized scores. The descriptor used a very high number of scores of the scale. Significant differences between both varieties were found in 13 out of 32 traits. Significant differences were found in 12 out of 32 traits between the Kladruby stud farm and the private studs. There is no reasonable explanation of this fact. A significant variety × stud interaction was recorded in only 7 traits. Significant differences between stallions and mares were recorded in a large number of front and body traits (in 11 out of 18 traits) and in 2 rear traits. Despite of a remarkable number of significant differences between the years of birth (in 18 out of 32 traits) and age at description (in 13 out of 32 traits) both factors are not the important ones for the selection of horses according to the type traits. The description of type traits, performance recording and selection are carried out in the population of four-years-old horses born in the given year. The age at description does not play a role because all horses are now described each year at a standard age of four years. The linear description of type is an important tool for the breeding and conservation of genetic diversity within the breed.