Does there exist a tabulation of the known rational homology of mapping class groups of genus $g$ with $m$ punctures? I'm most interested in the case when punctures can be permuted.
There's not a lot known about unstable cohomology. Here are a couple of known facts. I'll restrict myself to the closed case. 1) Of course, it known that the rational homology groups of the genus $1$ mapping class group all vanish. 2) Igusa proved that all the rational cohomology groups of the genus $2$ mapping class group vanish. I'm not entirely sure of the correct reference for this (I'm at home so I can't go look at my notes), but I think this is contained in his paper MR0114819 (22 #5637) Igusa, Junichi Arithmetic variety of moduli for genus two. Ann. of Math. (2) 72 1960 612–649. 3) The rational cohomology groups of the genus 3 mapping class group were calculated by Looijenga in MR1234266 (94i:14032) Looijenga, Eduard(NLUTREM) Cohomology of M3 and M13. (English summary) Mapping class groups and moduli spaces of Riemann surfaces (Göttingen, 1991/Seattle, WA, 1991), 205–228, Contemp. Math., 150, Amer. Math. Soc., Providence, RI, 1993. A correction to this is in Looijenga's paper "The Hodge polynomial of $\overline{M}_{3,1}$" with Getzler, which is available here. 4) Tommasi has calculated the rational cohomology groups of the genus $4$ mapping class group here : MR2134272 (2006c:14043) Tommasi, Orsola(NLRUNJ) Rational cohomology of the moduli space of genus 4 curves. (English summary) Compos. Math. 141 (2005), no. 2, 359–384. 5) The virtual cohomological dimension of the genus $g$ mapping class group is $4g5$. This implies that $H_k(Mod_g;\mathbb{Q})=0$ for $k > 4g5$. It is now known that in addition $H_{4g5}(Mod_g;\mathbb{Q})=0$ for $g$ at least $2$. This was independently discovered by two groups of people (Tom Church, myself, and Benson Farb are one group and Shigeyuki Morita, Takuya Sakasai, and Masaaki Suzuki are the other). Both groups have papers that will (hopefully) appear soon. John Harer has also claimed to have proven this in unpublished work. 6) From HarerZagier's computation of the Euler characteristic of the moduli space of curves, it follows that there exists exponentially many unstable classes! 


You should look at http://www.rasmusvillemoes.dk/thesis/thesis.pdf and M. Korkmaz' survey http://arxiv.org/abs/math/0307111 


A related MO question asks about Betti numbers of $M_{g,n}$ and has some quite complete answers. However, it focuses on the case where the $n$ marked points cannot be permuted. (Sorry, I know this should have been posted as a comment, but I do not have enough karma yet.) 

