The interpretation of mammograms is a complex perceptual cognitive skill involving the visual and interpretative evaluation of breast X-rays. In the context of the National Health Service Breast Cancer Screening Programme, reading mammograms entails distinguishing between large numbers of normal and few abnormal films, and making decisions with regard to further clinical investigation. This thesis is concerned with lay people learning to interpret screening mammograms. Three issues are considered in this context, the nature of mammogram interpretation skill, issues in training and individual differences. Previous research on radiological skill has involved the study of visual search, on the one hand, and analyses of cognitive expertise in radiology, on the other. In this thesis, it is argued that skill in screening mammogram interpretation involves both, visual analysis and interpretation of detected abnormality. This notion is formalised in a cognitive process model of mammogram interpretation, which incorporates two stages of processing. The model is supported by data from two skill acquisition studies, one longitudinal study of radiographers and one involving students who have received brief training. Performance outcome and error data from both studies confirm the two stage nature of mammogram interpretation skill. Issues in selection and training considered in the thesis reflect the applied concerns surrounding lay film interpretation in screening mammography. Individual differences in mammogram reading performance were investigated in terms of ability, personality and cognitive skills measures. Training was examined in terms of the effect of different instructional methods on interpretation performance. It was found that a feature oriented method of training lay people for screening mammogram interpretation has some advantages over more traditional training methods. Individual differences in film reading performance after training were best predicted by cognitive style and strategy variables.