Removing malicious nodes from any network is vital to its correct operation and enhances largely its security. This can be typically achieved by revoking or deactivating their credentials whenever a threshold of votes against them is reached. The problem of threshold based revocation schemes is that any colluding set of malicious nodes surpassing the threshold can revoke nodes in the networks at will. Lowering the threshold value, would make coalition attacks against the revocation protocol easier. Whereas increasing the threshold value would make the revocation of malicious nodes difficult. In this paper, we propose a new revocation scheme for mobile ad hoc networks which deals with the revocation problem adaptively, so that it allows the revocation of smaller sets of colluding malicious nodes easier, meanwhile it controls the number of erroneously revoked nodes caused by big sets of colluding attackers.