In this paper a new approach is taken to identify a crack in a simply supported plate undergoing free vibration. The approach uses a Markov-Chain Monte- Carlo implementation of Bayes' Rule to identify the presence of a crack and, more importantly, to estimate crack parameters; the process also provides confidence intervals for those parameters. To generate the required time series, a semi-analytical free response is constructed out of a finite element based eigen- solution. This detection technique is applied to a cracked plate and effectively identifies the crack location, orientation and length. The results show the utility and accuracy of this method for a variety of cracks lengths, suggesting that even small cracks may be detected.