In an extended series of papers by M. Leeder, J.R.L. Allen, J.S. Bridge, J. Alexander, and others, the depositional stacking patterns of channel-belt deposits were related to tectonic and other external controls (we refer to these models collectively as LAB models). These models established a basic connection between subsidence, channel avulsion, and the mesoscale (channel-belt) architecture of an alluvial deposit. Two fundamental predictions resulted from this work: (1) channel-belt stacking density and hence connectedness is inversely correlated to temporal (vertical) changes in sedimentation rate; and (2) channel-belt stacking density and hence connectedness is directly correlated to lateral changes in sedimentation rate. In this paper we make use of the Experimental EarthScape (XES) Facility at the St. Anthony Falls Laboratory to examine the effects of temporally and laterally variable subsidence rates on mesoscale alluvial architecture. We then compare the experimental results with the main predictions of the early LAB models. Regarding the predictions of the LAB models, under conditions of high sediment supply and a highly active alluvial system, lateral and downstream variation in subsidence geometry and rate have little effect on the details of alluvial architecture. We found that the principal architectural signature of changes in subsidence geometry (i.e., laterally variable subsidence rate) is stratal tilting and that channel belts are not attracted to the subsidence maximum. We hypothesize that the effect of variable subsidence geometry is felt by the fluvial system only if subsidence proceeds faster than the river can adjust to the formation of a topographic low via the deposition of overbank material in the form of splays and sheet sands. Our results lend impetus to the development of more complete three-dimensional numerical models of mesoscale alluvial architecture that couple architecture to broader, allogenic forcing mechanisms.