Based on the average sea surface temperature (SST)from May to September, the Schaefer model and the Fox model were used to estimate the catch for the autumn cohort of Todarodes pacificus.It is assumed that the ΔU (residuals of catch per unit effort between the observed and predicted values)calculated form the traditional surplus yield models were correlated with SST, and the SST-based surplus yield models were established. According to the overall development of the autumn cohort of T. pacificus fishery since the 1960s, the total catches from 1985 to 2014 were analyzed separately i.e.from 1985 to 1993, from 1994 to 2002, and from 2003 to 2014. According to with/without SST and whether to segment with SST, three Schaefer models and three Fox models were constructed respectively. Results indicated that the segmented Schaefer model-SST was the best, there was a negative linear correlation between ΔU and SST(P<0.05), and the catch decreased with the increase of temperature from 18 to 23℃. It is suggested that:the model establishment process should be segmented according to the different time periods, which can improve the model fitting; the segmental Schaefer model-SST can be used to evaluate the future catch to provide some reference for related resource management when calculating the maximum sustainable yield.