Superiority of genomic selection is due to the use of the realized relationships. Unfortunately, only a small portion of animals included in the evaluation are genotyped. Consequently, the majority of animals will still be evaluated based on their expected relationships. With the availability of genomic information, it is possible to assess the variation of expected co-ancestry coefficients for different orders (degrees) of relationships. Thus, this available information represents a good prior and could be used to better assess the Mendelian sampling in absence of genomic data. Furthermore, a sizeable portion of non-genotyped animals are already phenotyped for several traits with moderate to high heritabilities. Phenotypic records contain information about the true genetic relationship between animals. Using these 2 sources of information, we can improve the pedigree based relationship without using genomic data. A simulation was carried out to investigate the ability to infer realized additive relationships in absence of genomic information. A single trait with heritability of either 0.5 or 0.8 was simulated. A pedigree with large number of full (FS) and half (HS) sibs reflective of a chicken population was also simulated. All animals in the pedigree were assigned 60K SNP genotypes. The average (A) and the realized (G) matrices were computed using the pedigree and the genomic information, respectively. Using our procedure a realized relationship matrix (G*) was computed without use of the simulated genomic information. The results showed that the similarity between G and G*, measured by the absolute value of difference between corresponding elements of both matrices, was greater than between G and A. The average difference between G and A was 0.07 compared with 0.03 between G and G*. The realized relationships for FS and HS estimated using our proposed method overlap largely with those obtained using genomic information. Additionally, the similarity between G and G* increased, as expected, with the increase of heritability. The fact that the difference between G and G* is smaller than between G and A indicate that the matrix G* is a better approximation of G than A.