The author begins by underlining the ambiguity of the term 'temple treasures'. In fact, we only know the temples' art holdings from literary descriptions; no temple treasure has been found intact, particularly as a consequence of the suppression of pagan cult by Theodosius I. In every period temple treasures were pillaged (cf. Titus' triumph after the conquest of Jerusalem), while they also themselves included the products of pillage. Many coin hoards may have belonged to temples, without being identified as such; there were also private deposits placed under the protection of temples. Only finds in situ (for example in springs or river-sources) can definitely be attributed to divinities, but inscriptions can also help in identifying votive objects. To illustrate the make up of a Roman temple treasure, it is possible to compare the intact treasure of the Teng dynasty Buddhist temple excavated at Famasi, in Shansi province. The author then examines the objects from Roman temples found in Germania libera (fig. 1-3), then the many groups of objects found as a result of draining-work in the Palatinate, which certainly represent booty, of which some must come from temples (fig. 4-6). The reconstruction of temple treasures must take into account objects found in situ (votive offerings found in springs and streams, lararia, and scattered finds in temples) and the different kinds of treasures (non-utilitarian objects placed in favissae, deposits made in response to crisis or as a preventive measure, groups of stolen objects). The first category of undoubted cult objects is that of leaves and strips of metal, which are found unequally distributed and of various types (for example, in the form of 'naiskoi' in Egypt and North Africa), the signs and characteristic emblems of various cults. There follow the votive objects deposited in springs and streams (in France, often made of wood), which also include thousands of coins; these allow us to chart the development and disappearance of the cult. The author warns of the difficulty in interpreting fluvial deposits, where it is difficult to distinguish votive offerings from lost objects - particularly in the case of weapons. He then examines specific categories: decorations of lararia found in situ (preponderantly from the area of Vesuvius) or displaced (fig. 10), and objects from the cult of Jupiter Dolichenus (fig. 11). Discoveries from temple-sites, however, present similar problems of interpretation as fluvial deposits (fig. 12). The author analyses three great hoards from the limes, found at Weissenburg, Straubing and Mauer an der Url. The Weissenburg treasure, from its make-up, presumably came from a lararium (fig. 14). The Straubing hoard, very varied, seems to be loot. It is possible to distinguish, from the type and weight of the different metals, between