Adhesion is vital to composites because cracks are stopped at adhesive interfaces, deflected along them, or converted into interfacial dislocations. Failure is therefore dominated by the adhesion between the components of the composite. This paper describes new ideas about adhesion and considers the relevance of these ideas to composite materials. In particular it emphasises the difference between adhesion at the molecular level and adhesion in engineering terms. There is no doubt that molecules of solid materials are attracted to each other by van der Waals and other influences. Also, they can be separated by applying mechanical forces. The problem is explaining the connection, i.e. the mechanism, between molecular attractions and mechanical measurements. Ideas such as 'keying', 'glueing' and 'friction' require critical assessment because they operate differently at the molecular scale. Interesting mechanisms such as 'adhesive hysteresis', 'adhesive stringing', and 'adhesive aggregation' deserve evaluation. A rational theory of these and other interface phenomena should be based on the theoretical concept of reversible work of adhesion. However, it is necessary to distinguish this thermodynamic work of adhesion from the measured quantity of adhesive energy, which includes the extra energy required to restructure the interface as surfaces move.