Petrological collections result from sampling for academic research, for aesthetic or commercial reasons, and to document natural diversity. Selection criteria for reducing and enhancing collections include adequate documentation, potential for future use, information density, time and money invested in specimens, and spatial and financial constraints. Application of these criteria to the voluminous (c. 300,000 samples) rock collections of the University of Amsterdam, led to partial acquisition by the Nationaal Natuurhistorisch Museum in Leiden (Naturalis) late in 2002. Selected items included: (i) historical collections; (ii) material from former overseas domains; (iii) material from poorly accessible areas; (iv) material useful for research at the museum itself; (v) non-voluminous items with high information density (thin sections) or subjected to laboratory treatment (rock powders, mineral separates); and (vi) all samples quoted in academic dissertations. Promotion and advertising of the newly acquired collections is expected to lead to a second life for these important academic specimens. Application of similar criteria to other museum collections will lead to partial de-accessioning, thus creating space for future acquisitions in the framework where Naturalis is increasingly regarded as the Dutch national repository of geological collections. Researchers from partner institutions will be stimulated to (de-)select their collections at the end of a project, to avoid the much higher costs of later selection by museum staff.