The main objective of this study was to develop a new method to estimate relationship coefficients by combining molecular with pedigree data, which is useful for specific situations where neither pedigree nor molecular data are complete. The developed method was applied to contribute to the conservation of the Skyros pony breed, which consists of less than 200 individuals, divided into 3 main herds or subpopulations. In this study, relationships between individuals were estimated using traditional estimators as well as the newly developed method. For this purpose, 99 Skyros ponies were genotyped at 16 microsatellite loci. It appeared that the limitation of the most common molecular-based estimators is the use of weights that assume relationships equal to 0. The results showed that, as a consequence of this limitation, negative relationship values can be obtained in small inbred populations, for example. By contrast, the combined estimator gave no negative values. Using principal component analysis, the combined estimator also enabled a better graphic differentiation between the 3 subpopulations defined previously. In conclusion, this new estimator can be a promising alternative to traditionally used estimators, especially in inbred populations, with both incomplete pedigree and molecular information.