The performances of different sample sizes for estimating the size distribution of squat lobster (Pleuroncodes monodon) catches were analyzed using a computer resampling procedure. The data selected were gathered in May 2002 between 29 degrees 10' S and 32 degrees 10'S. These data were used to test seven sampling scenarios for fishing trips (1-7 trips), twelve scenarios of the number of individuals sampled per tow (25, 50,., 300), and two within-trip sampling strategies (sampling all tows and systematic tow sampling). By testing the combination of all these scenarios, we were able to analyze the performance of 168 scenarios of sample size for estimating the composition of sizes by sex. The results indicate a lower error index for estimates of the size frequency distribution as the number of fishing trips increases, with progressively smaller decreases among adjacent scenarios. Likewise, a decline is verified in the error index when increasing the number of sampled individuals, with minor improvements above 175 individuals per tow.