We consider the complexity of computing Boolean functions with algebraic decision trees over GF(2) and R. Some lower and upper bounds for algebraic decision trees of various degrees are found. It is shown that over GF(2) decision trees of degree d are more powerful than trees of degree < d. For the case of decision trees over R, it is shown that decision trees of degree 2 5 -0 (Jiilogo(')n) are more powerful than trees of degree en, with 0 < c < f .