This thesis details the work carried out under two research projects at the University of Surrey. The first project titled The Design of Structural Adhesive Joints', was of three years duration from September 1985 to August 1988 and was sponsored by the Science and Engineering Research Council. The second project, sponsored by Ford UK Ltd, and tided 'A General Joint Analysis Facility extended certain aspects of the analysis work initiated In the first period of research. The objective of the work was to address the problem of integrating structural adhesives Into the design process and to provide procedures that would facilitate this integration in a quantitative, rather than the more usual qualitative way. To be effective, such an approach needed to consider not only a means of analyzing a proposed joint but also a way of predicting the actual failure of that joint. An extensive literature survey of analyses available to the design engineer has been completed. The analyses investigated were found to be lacking in several critical respects, and as part of this research, methods of analysis overcoming some of these limitations have been developed. The analyses produced are based on earlier approaches but extended and modified as appropriate. The work on all the analyses produced has been carried out by considering a simple adherend-adhesive sandwich configuration. Five different analyses, considering the sandwich to be modelled with differing degrees of complexity, have been produced. In all of the analyses the adherends are assumed to behave as cylindrically bent plates capable of sustaining both tensile and shear forces and bending moments, with the adhesive transmitting both tensile and/or shear loads. Initially an elastic solution was obtained, adopting a relatively simple approach. This enabled the subsequent enhancement of including non-linear material behaviour to utilize the same governing equations, thus maintaining consistentcy. The General Elastic Analysis (GEA) has been extensively simplified to produce a number of two parameter design formulae suitable for use by an engineer at an early stage in the design process. The two analyses produced by this simplification are called the Simplified Peel Analysis (SPA) and the Simplified Shear Analysis (SSA), so called because they consider the named component of stress in the adhesive layer only. The GEA was then extended to include non-linear material properties in the adhesive layer, and an analysis called the Non-linear Adhesive Analysis (NLAA) was produced. A programme of validation using the NLAA and a non-linear finite element analysis of similar joint configurations was carried out. Additional comparisons with existing analyses have also been undertaken where possible. The NLAA has been shown to produce extremely accurate results for the stresses in the adhesive layer when compared with the component stresses predicted by the finite element method (FEM). The NLAA has been used successfully to determine the spread of yield in a single-lap joint, giving dose agreement with results from analysis using the FEM, but with much reduced computer and operator time. The final stage of the work was concerned with the Inclusion of non-linear adherend material properties, and an analysis called the Full Non-linear Analysis (FNLA) has been produced which Incorporates this refinement to the general model. Again the finite element method has been used to assess the accuracy of this new analysis, and the results from this work are presented here. Derivations of both forms of the elastic analysis and of the non-linear and full non-linear analyses are reported in Chapters 4 and 5 and the software appropriate to each Is described fully. The Initial survey of available literature has shown that there Is considerable lack of knowledge about possible causes of joint failure. Specifically, It Is noted that a criterion by which joint failure can be measured has not been uniquely defined. In an attempt to provide a criterion or criteria to enable the prediction of joint failure a 'Failure Criteria' test and analysis programme has been completed. Joint configurations were manufactured using a range of adhesives with different levels of ductility, and adherends of different stiffnesses. Batches of these test coupons were tested to failure under both predominantly mode I and mode 11ty pes of loading. Both FEM and FNLA analyses of each test configuration have been carried out, and the stress and strain distributions at the levels of failure load were established for each batch and studied to establish any correlation between various proposed failure criteria. Close agreement between certain factors and the equivalent bulk material properties was noted for test batches. The applicability of various failure criteria for both the mode I and mode II test configurations and possible general criteria are discussed. The failure of the mode I test configurations has been shown to be governed by the local level of maximum principal stress at the end of the overlap. The mode II test configurations also show dose agreement in terms of the maximum principal stress, but agreement with bulk data Is poor. Therefore, a further failure criterion is proposed for the mode II joints in terms of the 'global yielding' of the adhesive layer. The bulk property testing of the adherend and adhesive materials to establish their physical properties for use in the finite element analysis of the test programme Is also fully documented.