Fit quality?

Hello all,

when performing a MLL fit and storing the result in a RooFitResult,
what figure of merit should I use to compare the fit quality of two
fits on the same dataset?

Is fitResult->minNll() acceptable and if so, how can I compare it to
say a chi2/dof figure?
If not, is there a way (for binned MLL fits) to access a chi2 value?

Thanks in advance,
Laurent