Several studies have investigated the cognitive development of interacting peers. This study focuses on a phenomenon that has not yet been studied: the cognitive gains of 2 children with low levels of competence who fail to solve a task individually but who improve when working in peer interaction. We show that this phenomenon (which we call the two-wrongs-make-a-rightphenomenon) may occur when (a) the 2 wrongs disagree, (b) they have different strategies, and (c) active hypothesis testing is made possible. In a preliminary study, 30 Grade 10 low-achieving students were tested about the rules they use to compare 2 decimal fractions in a questionnaire. The students who were diagnosed as wrongs were invited to solve a task (the 6-cards task) with peers. Three kinds of pairs were formed: 7 W1-W2 pairs in which the 2 wrongs have different conceptual bugs; 4 W1-W1 pairs in which the 2 wrongs have the same conceptual bugs; 4 R-W pairs in which a wrong interacted with a right student. The 6-cards task was designed to create conflicts between students with different conceptual bugs and between wrong and right students. Two days after solving the 6-cards task, the students were asked to answer a similar questionnaire individually. The preliminary study revealed the two-wrongs-make-a-right phenomenon: Among the 7 W1-W2 pairs, at least 1 wrong became right. In contrast, in the 4 R-W pairs, only 1 wrong became right, and in the 3 W1 -W1 pairs, no change was detected. In a case study that replicated the phases of the preliminary study, disagreement, argumentative operations (such as challenge and concession), hypothesis testing (with a calculator), and the internalization of social interactions mediated the change of peers from wrongs to rights. We then replicated the initial study with 72 low-achieving Grade 10 and 11 students, confirming the two-wrongs-make-a-right effect.