Carbon molecular sieves are used in kinetically controlled separation processes. The mass transport in these materials can be quite complex as a result of their structure. In this contribution carbon dioxide uptake in a commercial carbon molecular sieve is analysed to show that the dynamic response does not correspond to a single mass transfer time constant. Experimental checks are discussed that ensure linearity and isothermal conditions as both have to be excluded in order to interpret correctly the dynamic curves. Surface barrier plus diffusion in micropores is compared to a distribution of surface barriers and the results show that a simple log-normal distribution of time constants reproduces with high accuracy the experimental transients. This is consistent with the way carbon molecular sieves are manufactured and requires only the definition of the variance of the distribution and the time constant at the mean.