A study examining damaging alkali-aggregate reaction (AAR) in concretes containing recycled aggregate is described. This considered masonry units, concrete, demolition waste, road planings, slate and plasterboard. Alkali release tests indicated that masonry units tended to release more alkali than recycled concrete (with higher levels generally obtained from recently produced materials), with slate and plasterboard giving least. Accelerated mortar expansion (reactivity) test results suggested there was a low risk of AAR for the materials. Concrete tests were then made using BS 812-123 exposure conditions. The concretes were proportioned (i) following the standard mixes. (Na2Oeq =7.0 kg/m2) with recycled materials replacing fine or coarse aggregate in a low-reactivity aggregate.combination and (ii) at high alkali level (Na2Oeq=5.4 kg/m3) with the materials as fine aggregate in a normalreactivity aggregate combination. The tests for (i) gave results for the recycled materials indicating low reactivity, with classifications of non-expansive or probably non-expansive to BRE Digest 330. For (ii), expansions for the materials were mainly similar/slightly higher than those found for (i). However, recycled concrete that previously. exhibited damaging AAR had greater expansions. Selected concretes were exposed at an external site for up to 8 5 years, during which observations by microscopy were carried out. These gave general agreement with the laboratory tests. The practical implications of the study are reviewed.