Abstract: In order to more objectively evaluate the technical efficiency of machinery manufacturing， two data envelopment analysis (DEA) models (both the optimal and the worst) were applied for the first time to estimate the technical efficiency of 85 listed machinery manufacturing companies of 2012， and by means of the geometric method the two envelopment frontiers results were reasonably sorted. The findings indicate that the optimal or the worst efficiency deviates from the combined efficiency in ranking， and the results are more objective when it comes to the combined efficiency; the fewer the input factors， the greater the volatility， or the more different the input variables are， the more differently the combined efficiency ranks as opposed to the optimal and the worst efficiency. Based on the above conclusions， listed companies are recommended to strengthen the balance of input factors and improve the efficiency of these factors.

Key words: technical efficiency, data envelopment analysis (DEA), optimal frontier, worst frontier, machinery manufacturing