Abstract: The length of test period of stationary performance testing (SPT) of stallions of the Dutch Warmblood riding horse population is evaluated. Heritability of successive judgements of traits and the phenotypic and genetic relations between successive judgements are estimated. Data from 1983-1988 are used, covering scores from 206 mostly 3-year-old stallions. Ten subjectively scored traits are considered: walk, trot, canter, riding ability, jumping ability, free jumping, cross country, character, stable behaviour, training report. Traits are successively scored at about 25, 50, 80 and 100 days in SPT. Missing scores are predicted on basis of the available scores using multiple partial regression coefficients. Validity of this method is checked in an independent data set for walk, trot and canter. The correlations between predicted and realized scores average 0.74, 0.77 and 0.79 when first, first and second, and first, second and third judgements are available, respectively. Variance and covariance components are estimated by restricted maximum likelihood (REML) procedures. Data from SPT are analysed using an animal model. Estimates of heritability are high and constant for gaits during the successive judgements. Except for stable behaviour estimates of heritability, the traits decrease slightly during the successive judgements. Estimates of the phenotypic and genetic correlations between successive judgements are high. It is concluded that length of SPT can be shortened and selection during SPT can be intensified.