I was looking up the Lennard-Jones parameter $\sigma$ (which, to my understanding, is a measure of molecular size) when I was surprised to discover that the value for the $\ce{N2}$ molecule (3.667 Å) was slightly lower than the value for the $\ce{CH4}$ molecule (3.78 Å). It's my understanding that $\ce{N2}$ molecules are significantly larger than $\ce{CH4}$ molecules, and this is exploited in, for instance, molecular sieves.
My question: is $\sigma$ a reasonable measure of mean atomic radius? And if it isn't, what is a good measure of molecular size, with which I can compare molecules?