Species richness is the most fundamental concept of diversity. The observed number of species in surveys are commonly used to represent the species richness in marine areas, however, this approach may lead to certain underestimations due to observational errors. This study used the marine fish community in Southern Shandong Peninsula as an example to test methods for reducing errors in species richness estimation. Non-parametric methods, Chao, Jackknife1, Jackknife2 and Bootstrap, were used to estimate the number of species in the waters of Southern Shandong Peninsula, based on bottom-trawl survey data collected in October 2016. In addition, we examined the factors that can influence the performance of non-parametric methods. The results show that the number of species in the community, estimated by four non-parametric methods, showed remarkable variations:265, 248, 235 and 206 for Jackknife2, Chao, Jackknife1 and Bootstrap, respectively. In addition, the four methods also showed differences in standard deviation, and the standard deviation was the largest in Chao, followed by Jackknife, and was lowest in Bootstrap. The results of MDS showed that fish assemblages could be divided into two communities and the number of species have large differences. The estimated values in community I were 250, 238, 219 and 191; the number of species estimated in community Ⅱ were 86 and 100. Comparative studies show that the sampling intensity and number of species affect the estimation of non-parametric methods. This study provides a method reference and comparison criterion for the comparison of species richness with different sampling efforts.