Advances in digital technologies for information capture combined with massive increases in the capacity of digital storage media mean that it is now possible to capture and store one’s entire life experiences in a Human Digital Memory (HDM). Information can be captured from a myriad of personal information devices including desktop computers, PDAs, digital cameras, video and audio recorders, and various sensors, including GPS, Bluetooth, and biometric devices. These diverse collections of personal information are potentially very valuable, but will only be so if significant information can be reliably retrieved from them. HDMs differ from traditional document collections for which existing search technologies have been developed since users may have poor recollection of contents or even the existence of stored items. Additionally HDM data is highly heterogeneous and unstructured, making it difficult to form search queries. We believe that a Personal Information Management (PIM) system which exploits the context of information capture, and potentially of earlier refinding, can be valuable in effective retrieval from an HDM. We report an investigation into how individuals perform searches of their personal information, and use the outcome of this study to develop an information retrieval (IR) framework for HDM search incorporating the context of document capture. We then describe the creation of a pilot HDM test collection, and initial experiments in retrieval from this collection. Results from these experiments indicate that use of context data can be significantly beneficial to increasing the efficient retrieval of partially recalled items from an HDM.