We present a study into the use of smell for searching digi-tal photo collections. Many people now have large photo libraries on their computers and effective search tools are needed. Smell has a strong link to memory and emotion so may be a good way to cue recall when searching. Our study compared text and smell based tagging. For the first stage we generated a set of smell and tag names from user de-scriptions of photos, participants then used these to tag pho-tos, returning two weeks later to answer questions on their photos. Results showed that participants could tag effec-tively with text labels, as this is a common and familiar task. Performance with smells was lower but participants performed significantly above chance, with some partici-pants using smells well. This suggests that smell has poten-tial. Results also showed that some smells were consistently identified and useful, but some were not and highlighted issues with smell delivery devices. We also discuss some practical issues of using smell for interaction.