I have compared dijet cross-sections calculated using code from Daniel de Florian and calculations made by Markus Wobisch using FastNLO.  For the de Florian code, I have calculations at LO (using MRST 2001 and CTEQ6L pdfs) and NLO (using MRST2004, CTEQ6M, and CTEQ6.6 pdfs) as well as an LO calculation using the NLO CTEQ6.6 pdf.  For FastNLO I have only LO and NLO calculations with CTEQ6.6.  Note that green error bars are scale errors.

FastNLO-de Florian comparison for NLO, with ratio (ratio calculated for de Florian with CTEQ6M)

Comparison for LO, with ratio (ratio calculated for de Florian with CTEQ6L)

Comparison of LO calculations with NLO pdf CTEQ6.6 (in the top plot, red circles are FastNLO, blue triangles de Florian)

Comparison of LO calculations with NLO pdf CTEQ6.6 (in the top plot, red circles are FastNLO, blue triangles de Florian)