Free-access Legal Information Institutes (LIIs) like AustLII have an increasing wealth and diversity of legal data, including legislation from numerous jurisdictions, decisions of both general and specialised tribunals, and sometimes law reform reports, law journals, treaties etc. This profusion of content leads to problems in precision of searches. One way commercial publishers have dealt with this, and added value to their content, is by creating subject-specific research facilities on topics such as environmental law, IP, criminal procedure etc. The challenge for free-access LIIs is that any such value-adding cannot involve the high costs of constant editorial intervention, nor the commissioning of subject-specific commentaries. This paper explains experiments by AustLII to create useful subject-specific 'Libraries' in areas of Australian law such as indigenous law, taxation and industrial law, and on international and humanitarian law (on WorldLII). The experiments involve methods of identifying and isolating within databases of general content that which is on specified subjects, by largely automated and repeatable means, particularly the use of approximating searches. Methods of testing the searches used to construct Libraries are suggested. Once a subject-specific Library is created, it can be used as a context to provide access to equally specific content provided by search engines or commercial publishers. Systems like WorldLII which draw together the content of many LIIs pose similar challenges at the international level, but may make possible international comparative law facilities.