Several production systems, mainly for pig and chicken, are based on crossbreeding. However, one limitation of associated breeding programs is that selection is performed on purebreds while the aim is to improve crossbred performance. Also, in crossbred populations, single nucleotide polymorphism (SNP) effects may differ between purebred and crossbred populations due to different genetic backgrounds and environments, and, thus, may be line (breed) specific. Assuming that purebred origin of alleles in crossbreds is known, different implementations for genomic selection of purebreds for crossbred performance were recently proposed. Results from simulations showed that genomic models that consider line (breed) specific SNP effects could outperform current genomic models that assume the same SNP effect across breeds. Estimation of breed specific SNP effects in real datasets will require phasing of genotypes in crossbreds and assigning breed origin of their alleles. Therefore, an approach was developed that applies long range phasing to assign the purebred origin of alleles of crossbreds. Long range phasing methods can be applied in the absence of pedigree and with the presence of haplotypes within a Linkage Disequilibrium block common between breeds. The aim of this study was to determine the accuracy of allelic assignment of the proposed approach on datasets simulating crosses between closely related, distantly related and unrelated breeds. Effects of different factors and parameters, like the presence of a haplotype in another pure breed that would preclude the assignment to the first pure breed, were also tested. An application on real pig genotype data was performed. Determining the allelic origin of crossbreds was feasible without pedigree information.