Classification techniques can be used as an alternative to the selection index, especially for indirect selection of multiple traits. Prior information for this technique provides to increase accuracy of selection. Two sets of simulations were used to estimate the error rates of the classification techniques in selection of livestock and to compare the method with the animal model, an example of Best Linear Unbiased Predictions (BLUP). The first set of analyses considered the traits milk yield, milk fat, milk protein and somatic cell count (SCC). Frequent milking was simulated and the high and low means were created within the standard deviation limits. Using these different means, observations and standard deviations, various scenarios were produced to simulate various data sets for validation to check how correctly the classification method worked. The different scenarios included A. Same means and sigma, different observations, B. Different means and observations, same sigma, C. Different means (standardized) and observations, same sigma, D. Same means, different sigma (+/- 5 for milk yield, milk fat and protein, +/- 3 for SCC) and observations, E. Same means, different sigma (+/- 10 for milk yield, milk fat and protein, +/- 6 for SCC) and observations. The procedure selected the correct animals and produced a very low error rate in almost all simulations. In the second set of simulations, the traits considered were milk yield and SCC for two goat breeds, Saanen and German Fawn. A total of 10 000 dams, 3174 sires and 10 000 progeny were simulated. The family variance was 4.5. The sires had 3.15 progeny on average, and sires had minimum 1 and maximum 15 progeny. Estimated breeding values from the MTDFREML (Multi Trait Derivative Free Estimated Maximum Likelihood) program were used in the culling decisions. Animals selected using the animal model of BLUP with MTDFREML and animals selected using the classification method with SAS were analyzed employing the rank correlation, and the calculated value was very high 0.99 (P<0.01). If used with the necessary standardizations, the method seems promising enough to be considered as an alternative to existing selection methods and a new method for indirect, multiple trait selection in animal breeding. In addition, the method can be used in cases (no extensive pedigree, small number of animals/sires etc.) where it is difficult to calculate correct genetic parameters.