The solar chromosphere is a very dynamic and highly structured environment. In contrast to the higher density hydrodynamically controlled photosphere, this environment is controlled by magnetohydrodynamic forces. The lower density chromosphere encompasses the transition from optically thick to thin so that Non-LTE radiative transfer applies. While the omni-presence of emission in UV lines indicates that a chromospheric rise in temperature from the cool photosphere to the hot corona should occur everywhere on the solar surface, the presence of cool material at chromospheric heights observed in infrared CO lines has made us reconsider the validity of using this interpretation of the observed spectrum based on average one-dimensional models for chromospheric thermal structure. It can be shown that the mean thermal properties of spatially inhomogeneous and/or time-dependent atmospheres in fact are not well represented by an average one-dimensional semi-empirical model based on the mean spectrum. Observations of CO lines show that the coolest locations in the upper photosphere/lower chromosphere are associated with the strong adiabatic expansion that occurs over granules. Simulations that try to model chromospheric dynamics should therefore probably include the effects of solar convection, wave dynamics, as well as the effect of radiative cooling in the multitude of infrared CO lines on the thermal balance of the atmosphere.