Many 40Ar/39Ar age spectra for alkali feldspars are significantly different from the model age spectra calculated for slowly cooled samples composed of diffusion domains of a single size, and the Arrhenius plots for these samples show departures from linearity that are inconsistent with diffusion from domains of equal size. The most plausible explanation for these discrepancies is the existence of a distribution of diffusion domain sizes. We have extended the single-diffusion-domain closure model of Dodson so that it applies to minerals with a distribution of domain sizes and have used it to explain many commonly observed features of 40Ar/39Ar age spectra and Arrhenius plots for 39Ar loss during step heating. For samples with a distribution of diffusion domain sizes, the form of the 39Ar Arrhenius curve is a function of the heating schedule (i.e., the temperature and duration of the steps used), and thus different heating schedules will result in different curves for the same sample. This effect can be used to confirm the existence of a distribution of diffusion domain sizes and to optimize the information contained in the Arrhenius plot. The multiple diffusion domain size model is used to reinterpret the age spectra, Arrhenius plots, and cooling history of three feldspars from the Chain of Ponds pluton, northwestern Maine, earlier interpreted assuming a single domain size. Interpreting the 40Ar/39Ar and 39Ar released during step heating in terms of a single domain size gives rise to a large discrepancy between the cooling rate determined from the age and closure temperature of the three samples compared to the cooling rate required to explain the shape of the individual age spectra. The single domain size model fails also to account for the observed departures from linearity of the Arrhenius plots. We show that a particular domain size distribution in each sample can explain in detail both the shape of the age spectra and the Arrhenius plots, and results in the three samples defining a common cooling history. There is thus good evidence for the three alkali feldspar samples studied here that the thermally activated diffusion measured by 39Ar release during step heating in the laboratory is also the mechanism responsible for argon loss or retention in the natural setting.