The behaviour of Group I and II elements during the petrogenesis of felsic igneous rocks is largely controlled by feldspar-liquid relationships. Numerous experimental studies have addressed plagioclase/melt element partitioning, with fewer studies devoted to potassium feldspar, and very few to albite-rich ternary-composition feldspar (An ∼ Or < Ab). However, the partition coefficient for Ba is known to increase by at least an order of magnitude through the crystallisation sequence sodic plagioclase–anorthoclase–potassium feldspar that is typical of sodic alkaline suites. Feldspars, glasses, and whole rocks in such suites may exhibit strong enrichments and depletions that can be used to track processes of crystal fractionation, cumulate formation, and cumulate recycling. Here, we review experimental feldspar/melt partitioning data for Ba, Sr, and Rb for all feldspars. Regression of available data provides expressions that appear to adequately model the compositional and temperature dependence of partition coefficients for albite-rich compositions. We have applied this model to feldspar and melt compositions of the products of several Holocene eruptions (Pico Viejo C, Pico Viejo H, Teide J2, Lavas Negras, Arenas Blancas, Montaña Rajada and Montaña Reventada) of the basanitic-phonolitic suite of the Teide-Pico Viejo volcanic system (Tenerife, Spain), using EPMA and LA-ICP-MS analyses. Comparing analysed feldspar/groundmass pairs with predicted partition coefficients obtained with the models provides a way of distinguishing between feldspars that are in or out of equilibrium with their host melt, and of reconstructing feldspar histories. The results demonstrate the existence of a distinct population of feldspars that had undergone accumulation, fusion and recrystallisation events, in Lavas Negras and Arenas Blancas flows. In addition, the anomalous trachytic composition of Montaña Reventada is due to melting of a feldspar-dominated cumulate. Application of these techniques to active magmatic systems will allow us a better understanding of different pre-eruptive processes, and ultimately improve volcanic hazard assessment.