Let be an array of symbols. is called avoidable if for every set of symbols, there is an Latin square on these symbols so that corresponding cells in and differ. Due to recent work of Cavenagh and Ohman, we now know that all partial Latin squares are avoidable for . Cavenagh and Ohman have shown that partial Latin squares of order for [1] and for [2] are avoidable. We give a short argument that includes all partial Latin squares of these orders of at least . We then ask the following question: given an partial Latin square with some specified structure, is there an Latin square of the same structure for which avoids ? We answer this question in the context of generalized sudoku squares.