This review explores the potential for establishing valley incision chronologies from alluviumfilled cave systems, and covers a total of 30 case studies since 1997. Caves in limestone develop very fast (~10^4 years) when conditions for bedrock solution are optimal, and many contain alluvium deposited by allogenic sinking streams, preserving the sediment thereafter for millions of years. Cave networks display a vertical succession of sub-horizontal passages which indicate past positions of the water table, with the stream in-and outlet caves indicating the former elevation of the adjacent valley floor. Abandoned cave levels are expected to multiply as valley incision increases local relief (descending speleogenesis), but sediment aggradation or glacier ice accumulation may also raise the local base level and flood older caves or generate new ones (ascending speleogenesis). Establishing the age of alluvial sediment hosted by caves relies on burial dating of quartz-rich clasts using two terrestrial cosmogenic nuclides (TCNs) -commonly 26Al and 10Be -measured in the same sample. Systematic examination of age-elevation data patterns in the existing literature reveals situations ranging from intuitively consistent valley incision histories to counter-intuitive age inversions and other anomalies. Here those anomalies are analyzed and classified in order to establish the extent to which the corresponding inconsistencies are avoidable, thereby providing a methodical catalogue of foreseeable difficulties and pitfalls. Three domains of uncertainty are emphasized. The first relates to karst processes: cave network geometry, cave passage response to vadose and phreatic processes, and diachronous links between cavity age and sediment. The multiple pathways of speleogenesis are reviewed. They highlight ambiguities behind the concept of 'cave level', which, as a proxy for baselevel paleoelevations, may be less precise than subaerial information provided by fluvial fill or strath terraces. The second source of uncertainty lies in the chronological information provided by the alluvium. Sediment dynamics in subterranean karst generate complicated stratigraphic configurations, with opportunities for postdepositional sediment reworking within or between cave levels. Furthermore, a TCN burial age is valid for a population of quartz grains but not necessarily for the entire stratigraphic sequence containing them nor for the cave that contains it. The third source of uncertainty lies in the burial dating method itself, because 26Al and 10Be nuclide inventories cannot unequivocally document whether older burial events might have occurred prior to final burial in the cave. The review recommends that (i) sampling strategies should be contingent on a diagnosis of speleogens and speleothems, and on a detailed sedimentological and stratigraphic analysis of the alluvial fills; (ii) dating should focus on individual bedload clasts rather than on sand because this helps to discriminate between pebble populations and to detect sediment mixing; (iii) 26Al/ 10Be ratios in modern channel alluvium and in older deposits stored in the catchment should be measured for the purpose of detecting whether certain features endemic to the sediment cascade could explain apparent burial age anomalies in the cave sediments. In situations where 26Al/ 10Be determinations generate wide age dispersion, four scenarios are discussed in which either the oldest or the youngest age should be retained.