Alkali feldspar 40Ar/39Ar and apatite UPb geochronological studies have typically invoked two mechanisms to account for apparent loss of radiogenic 40Ar and 206-208Pb. Some studies have suggested that the radiogenic isotopes were lost by volume diffusion and used these dates to constrain temporal variations of rock temperatures; others have argued that the radiogenic isotopes were lost due to interaction with fluids and related these dates to chemical alteration. These two end-member interpretations have fundamentally different implications for tectonic models derived from geochronological data, and therefore it is important to reliably identify the principal mechanism for loss of radiogenic isotopes. Here, we revisit the mechanisms of 40Ar loss in the famous gem-quality alkali feldspar from the Itrongay pegmatite in Madagascar. Previous studies have suggested that volume diffusion is the dominant mechanism of 40Ar loss, providing key evidence to support the use of 40Ar/39Ar dating of alkali feldspar for thermochronology. We attempted to verify these results by obtaining time-temperature paths from petrologically characterised cogenetic feldspar and apatite from the Itrongay pegmatite and comparing them with each other. However, our results suggest that only a minor component in the variability of 40Ar/39Ar dates of Itrongay feldspar is related to the diffusive loss of 40Ar, and that this loss was not compatible with the majority of previously proposed models, which hinders quantitative interpretations. The crystal studied here grew in five episodes related to the influx initially of co-existing dense SiO2-rich solution and CO2-dominated fluid (the first and supposedly the following two episodes) and subsequently of H2O-rich fluid (supposedly the final two episodes). Much greater component in the variability of the acquired 40Ar/39Ar dates is interpreted to reflect the differences in the ages of these growth episodes, which we estimate to span from 477 Ma to 176 Ma (the first four episodes). Apatite inclusions in this crystal are interpreted to be xenocrysts derived from the country rocks of the Itrongay pegmatite. These yield older UPb dates than the estimated age of their host feldspar and have apparently experienced diffusive loss of 206,207Pb prior to entrapment. Our 40Ar/39Ar results indicate that there is a lack of unambiguous evidence for diffusive loss of 40Ar from alkali feldspar that can be readily interpreted for thermochronological purposes. However, in situ 40Ar/39Ar dating of alkali feldspar appears to be a promising tool for tracking fluid-flow events in the Earth's crust whose applicability is not restricted to sedimentary rocks. Our UPb results corroborate previous suggestions that UPb dating of apatite can be used for thermochronology.