Many estimators and algorithms have been developed to infer the genealogical relationships from molecular marker data when there is no pedigree information. Most pairwise methods provide estimates in a continuous range that needs to be converted into genealogical relationships (namely full-sibs, half-sibs and unrelated) if there is a previous knowledge of the population structure. Transformations are usually based on arbitrary thresholds, but the possibility of correcting the coancestry estimates via explicit pedigree reconstruction methods has been suggested. Using molecular data for ten highly polymorphic microsatellite loci on a population of turbot (Scophthalmus maximus) with a known genealogy, the behaviour of four pairwise marker-based coancestry estimators and the molecular coancestry has been evaluated. The population consisted on 138 families with 4 full-sib individuals each and one family with 8 full-sib individuals (up to 15 half-sib families were also present due to the sharing of parents between some of the full-sibs families). Our results suggested that transforming the pairwise estimators and the molecular coancestry to family relationships through the establishment of thresholds performs slightly better than the explicit pedigree reconstruction method, when accuracy is measured in a pairwise basis. However, if joint relationships between more than two individuals were tested at a time, the threshold methods led to a high percentage of incongruous triads of full-sib individuals, with Mendelian incompatibilities appearing in congruous full-sib families (more than 70% and 60% in our study, respectively). The explicit pedigree reconstruction approach, due to its nature, is free from such problems. Therefore, the pedigree reconstruction approach seems to be a valuable tool to provide a congruent and compatible pedigree when the pairwise marker-based coancestry matrices or the molecular coancestry need to be transformed. © 2007 Elsevier B.V. All rights reserved.