This paper presents a new likelihood method for inferring full- and half-sibships from genetic marker data, incorporating models of typing errors to improve accuracy. The method addresses the issue that traditional likelihood methods assume perfect marker data, which is rarely the case in practice. Simulations show that ignoring typing errors can severely bias sibship estimates, but the new method can accurately infer sibships even with high error rates and identify typing errors at each locus. The method also improves upon previous approaches by using an iterative procedure to update allele frequencies, allowing for parental information, and using efficient algorithms for likelihood calculations. It is tested on simulated data with varying numbers of marker loci, error rates, and family structures, and applied to two empirical data sets. The method accounts for both class I (allelic dropout) and class II (stochastic) typing errors, and uses Bayes' theorem to estimate allele frequencies and infer parental genotypes. The performance of the method is evaluated through simulations, showing that it is robust to violations of assumptions and provides accurate sibship inference even with high error rates. The results demonstrate that accounting for typing errors is crucial for accurate sibship inference, particularly in haplo-diploid species where typing errors have a greater impact. The method is shown to be effective in identifying typing errors and inferring parental genotypes, and its accuracy is measured using statistics such as the proportion of correctly inferred full-sib pairs and the proportion of correctly detected typing errors. The results highlight the importance of incorporating typing error models in sibship reconstruction to ensure accurate inference of genetic relationships.This paper presents a new likelihood method for inferring full- and half-sibships from genetic marker data, incorporating models of typing errors to improve accuracy. The method addresses the issue that traditional likelihood methods assume perfect marker data, which is rarely the case in practice. Simulations show that ignoring typing errors can severely bias sibship estimates, but the new method can accurately infer sibships even with high error rates and identify typing errors at each locus. The method also improves upon previous approaches by using an iterative procedure to update allele frequencies, allowing for parental information, and using efficient algorithms for likelihood calculations. It is tested on simulated data with varying numbers of marker loci, error rates, and family structures, and applied to two empirical data sets. The method accounts for both class I (allelic dropout) and class II (stochastic) typing errors, and uses Bayes' theorem to estimate allele frequencies and infer parental genotypes. The performance of the method is evaluated through simulations, showing that it is robust to violations of assumptions and provides accurate sibship inference even with high error rates. The results demonstrate that accounting for typing errors is crucial for accurate sibship inference, particularly in haplo-diploid species where typing errors have a greater impact. The method is shown to be effective in identifying typing errors and inferring parental genotypes, and its accuracy is measured using statistics such as the proportion of correctly inferred full-sib pairs and the proportion of correctly detected typing errors. The results highlight the importance of incorporating typing error models in sibship reconstruction to ensure accurate inference of genetic relationships.