Serial correlation in data collected from fisheries acoustic surveys may have an effect on the precision and accuracy of fish abundance estimates. A simple random sample approach to the data analysis yields unreliable confidence intervals for mean population density when the degree of serial correlation in the data is high. The results of a simulation analysis indicate that more reliable confidence intervals can be obtained using cluster sampling estimation techniques.Key words: acoustic surveys, fish abundance estimates, cluster sampling, serial correlation