In this paper, a system for understanding mathematical expressions ispresented. After separating all symbols in an input mathematical expression, we utilize thirteen features to represent each symbol. In order to reduce the computational time, we apply a coarse classification algorithm to reduce the number of candidates. Then for each input symbol, we select the character with the highest similarity as the candidate symbol. Since some of the symbols in an arithmetical expression may touch each other, a dynamic programming algorithm is adopted to identify correct characters from connected symbols. In the expression formation stage, we propose a procedure-oriented method to translate the recognized symbols appearing in a 2-0 space into a 1 -D character string. We have used 105 mathematical expressions as training data and 50 expressions as testing data. The experimental results have demonstrated the feasibility of the understanding system.