We examine the effect that the shape of the source brightness profile has on the magnitude fluctuations of images in quasar lens systems due to microlensing. We do this by convolving a variety of accretion disk models (including Gaussian disks, uniform disks, "cones," and a Shakura-Sunyaev thermal model) with two magnification maps in the source plane, one with convergence κ = 0.4 and shear γ = 0.4 (positive parity) and the other with κ = γ = 0.6 (negative parity). By looking at magnification histograms of the convolutions and using χ² tests to determine the number of observations that would be necessary to distinguish histograms associated with different disk models, we find that, for circular disk models, the microlensing fluctuations are relatively insensitive to all properties of the models except the half-light radius of the disk. Shakura-Sunyaev models are sufficiently well constrained by observed quasar properties that we can estimate the half-light radius at optical wavelengths for a typical quasar. If Shakura-Sunyaev models are appropriate, the half-light radii are very much smaller than the Einstein rings of intervening stars, and the quasar can be reasonably taken to be a point source except in the immediate vicinity of caustic-crossing events.