A method for estimating mean catch per unit effort (c.p.u.e.) with groundfish trawl survey data is proposed and examined in detail. The new method is a modificaiton of the arithmetic average of the survey c.p.u.e.s. It is meant to be applied to those cases where one or more of the survey c.p.u.e.s are excessively large and exert too much influence on the arithmetic average, when it is used to estimate mean catch per unit effort. These large c.p.u.e.s are replaced by estimates of their expected values.