A diatom-salinity transfer function is developed for quantitative palaeosalinity reconstruction. from a data-set of 74 modem diatom samples and limnological data from Spanish salt lakes. Relationships between species and environmental data are explored using principal components analysis (PCA), detrended correspondence analysis (DCA) and canonical correspondence analysis (CCA) with variance partitioning. and a conductivity transfer function is derived by weighted averaging (apparent r = 0.91). A model for direct lake-level reconstruction is developed using lWINSPAN to explore the relationship between diatom species distribution and lake permanence. A survey is made of diatom preservation in the recent sediments of >50 lakes throughout Spain. and of the overall potential for Holocene reconstruction of a selected group of lakes. Diatoms are best preserved (least dissolved) in systems of southern Spain. and the Laguna de Medina is selected for analysis as a result. It is established by PCA that the correlated variables water depth and lake permanence. and salinity. are the most important factors influencing preservation. The diatom record preserved in the lower 6m of a lO.3m core from the Laguna de Medina is analysed. Application of the transfer function is shown to be unreliable due to the effects of fossil diatom dissolution and a lack of modem analogues. A modified transfer function is derived which includes partially-dissolved assemblages and improves the match with the surface set. but remains unreliable. Palaeosalinity reconstruction therefore relies on traditional palaeoecological techniques. Lake-level change is also reconstructed independently from changes in the relative abundance of taxa Separate palaeosalinity and lake-level reconstructions are derived from lithological and ostracod shellchemistry analysis. and the palaeoecology of ostracods, foraminifera, molluscs and plant macrofossils. The influence of human impact is assessed by comparison with the pollen and charcoal record. The diatom record is strong despite dissolution effects, and a consistent interpretation of lake-level change can be made based on different indicators. Lakes were high prior to c. 7860 yr BP; following a phase of low. fluctuating lake levels. maximum lake level was attained around 6000 yr BP. after which it declined gradually. The record shows affmities with African and Mediterranean lake-level data and makes an important contribution to an understanding of Holocene climate change in these regions.