The problem of detecting damage in composite plates is addressed here using L amb waves and novelty detection. Damage can be inferred from the scattering and modification of the Lamb wavefield as it passes through a defect. In order to produce an automatic diagnostic tool which can operate on measured time data, the method of novelty detection is used. This depends on establishing a description of normality which then allows subsequent signals to be flagged as anomalous if they deviate from normal condition. Three methods of novelty detection are illustrated: two statistical methods and one neural. The methods are demonstrated on experimental data captured from two composite plates.