This paper considers a lower bound estimation over risk for d dimensional regression functions in Besov spaces based on biased data. We provide the best possible lower bound up to a lnn factor by using wavelet methods. When the weight function ω(x,y)≡1 and 3252b" title="Click to view the MathML source">d=1, our result reduces to Chesneau’s theorem, see Chesneau (2007).