An empirical approach was used to determine the sample size of transects in acoustic surveys to estimate the abundance of three pelagic species in northern Chile. Relationships between the coefficient of variation of fish density and modified degree of coverage were established, where the modified degree of coverage is proportional to the distance sailed and the surface occupation index (Co) of the species and inversely related to the square root of the study area. From this relationship, an equation was obtained to estimate the number of transects required in order to obtain a predefined level of precision, given a known level of occupation. The surface occupation index corrects the degree of coverage and explains to a large degree the differences in the estimated coefficients of variation among the different species. Results showed that sample size declined with an increase in the surface occupation index of the species, and the magnitude of that reduction was appreciably greater for higher levels of precision. The number of transects must be limited to sample sizes with a minimum transect separation in order to assure independence between transect densities. The empirical procedure used for the estimation of the number of transects can be applied to other species situations if the information is available from previous surveys, since the approach only requires repeated echo integrator surveys.