Morphological measurements of Lipizzan horses from state-owned studs in seven countries in Central and South-east Europe (Austria, Croatia, Hungary, Italy, Romania, Slovakia and Slovenia) were recorded to characterise the base population of Lipizzan horse breeding. A total of 37 distance and angular measures were taken from 368 breeding mares and 145 stallions from eight studs tone stud from each country with the exception of Romania, where horses were measured at two studs), Significant differences between studs were found for most of the traits (34 of 37 for mares, 19 of 37 for stallions), Repeatabilities of measurements (based on two repetitions in 100 horses) varied from 0.23 to 0.95. Some of the measurements most frequently taken in horse breeding showed high repeatabilities (0.95 for height at withers and cannon bone circumference), but chest girth showed a rather low repeatability of 0.45. Heritabilites of the traits Varied from 0.00 to 0.55. Estimates were comparable to literature results but due to the small data set (only the 368 mares were used for analysis) they were partly inconsistent. Multivariate analyses were applied separately to mares and stallions. Length of neck, cannon bone circumference, and width of thorax were the most important traits for discriminating studs for mares: width of thurls, chest circumference, and width of hips were important for stallions. Mahalanobis distances based on all 37 measurements were significant for pairwise comparison of all studs. An analysis of the differences of sire lines showed no significance. The results reflect at least partially different breeding goals. (C) 2001 Elsevier Science B.V. All rights reserved.