Ropalidia marginata is a social wasp in which colonies consist of a single fertile queen and several sterile workers. If the queen is removed, one of the workers, potential queen (PQ), becomes hyperaggressive and becomes the next queen. The identity of the PQ cannot be predicted in the presence of the queen. The probability of a worker succeeding the queen is uncorrelated with herbody size, dominance rank, ovarian or mating status, but imperfectly correlated with her age. Here, we investigate whether genetic relatedness help to predict the queen’s successors. We constructed models based on successors being (i) most closely related to the queen, (ii) most closely related to the immediate predecessor queen/PQ, or (iii) having the highest relatedness to the majority of theworkers; and (iv) having the highest average relatedness to all theworkers.We predicted five successors fromeach of these models using pair-wise genetic relatedness estimated from polymorphic microsatellite loci. We independently performed serial queen/PQ removal experiments and compared the observed sequence of successors with the predictions from the models. The predictions of none of the models matched the experimental results; on an average 5-6 individuals predicted by the models were bypassed in the experiment.Thus, genetic relatedness is inadequate to predict the queen’s successors in this species.We discuss why relatedness sometimes predicts the patterns of altruistic behaviour and sometimes not, and argue that the cost and benefit terms in Hamilton’s rule, i.e. ecology, should be vigorously investigated when relatedness does not have adequate explanatory power.