Line Spectral Frequencies (LSFs) are often used as parameters to represent the vocal tract filter in speech coders using linear prediction. We propose a new method for the quantization of the LSFs, namely Predictive Delta Adaptive Scalar Quantization (PDASQ). This method is a scalar quantization scheme based on a non-linear two-dimensional prediction in the index domain. It is shown that it can be implemented efficiently with negligible computational overhead and memory requirements compared to the simple scalar quantization method. Despite its lower bit rates the quantization distortion resulting from PDASQ is of the same order as with conventional scalar quantization. Satisfactory performance of the new method is verified through experimental tests using computer simulation.