We consider using microsatellites for paternity checking and parent identification in different population structures, and allowing for possible typing errors or mutations. Statistical rules derived from the Bayesian and the sampling approaches are discussed in the case involving the choice of the true father–mother pair among a finite set of possible parental pairs. General situations are investigated by means of random simulations, in order to characterize the joint influences of the number and polymorphism of typed loci, the population structure and size, and error rates. Approximate expressions are provided that give the efficiency of a set of markers for identifying the parents in various mating schemes. The importance of a non-zero value for the typing error rate in the likelihood is highlighted.