Abstract This study uses different morphometric methods to study morphological variation in the population of Engyprosopon grandisquama near the southwestern coast of Taiwan. In addition to the traditional multivariates method , geomorphmetric methods, including the Baselined method and General resistant fit method were used and the results were compared. Results from ocular-side and blind-side analysis are similar. Appearance among locations is different. Overall, each sample including all specimens from all stations can be classified into either two or three groups, depending on the employed method, as well as characters of blind and ocular sides being used. The groups are¡GYung-an, Tso-ying, and Kao-ping-hsi. The latter area appeared to be shifting between the other two groups. Bigger fish, estaminated by centroid-size, morphometric methods, appeared to stay in deeper water(60~80m), and their morphometric performance is different from those in shallow water. Analysis of variance from the General linear model showed significant regression relationships of centroid-size against distances of sampling site to different harbor estuary. The orders of their effectiveness of dosal-side are Tso-ying, Kaohsiung, Yung-an and Kao-ping-hsi respectively. Effectiveness based on blind-side characters are Tso-ying, Kaohsiung, Kao-ping-hsi and water depth. Generally speaking, most samples can be separated into three area-groups, which are¡G(1)Yung-an, (2)Tso-ying and (3)Kao-ping-hsi. Effect of water depth are better expressed by the blind-side characters, whereas effect of the area-sites are better expressed by the ocular-side characters.