Empirical studies using visual search methods to investigate spider communities were conducted with different sampling protocols, including a variety of plot sizes, sampling efforts, and diurnal periods for sampling. We sampled 11 plots ranging in size from 5 by 10 m to 5 by 60 m. In each plot, we computed the total number of species detected every 10 min during 1 hr during the daytime and during the nighttime (0630 hours to 1100 hours, both a.m. and p.m.). We measured the influence of time effort on the measurement of species richness by comparing the curves produced by sample-based rarefaction and species richness estimation (first-order jackknife). We used a general linear model with repeated measures to assess whether the phase of the day during which sampling occurred and the differences in the plot lengths influenced the number of species observed and the number of species estimated. To measure the differences in species composition between the phases of the day, we used a multiresponse permutation procedure and a graphical representation based on nonmetric multidimensional scaling. After 50 min of sampling, we noted a decreased rate of species accumulation and a tendency of the estimated richness curves to reach an asymptote. We did not detect an effect of plot size on the number of species sampled. However, differences in observed species richness and species composition were found between phases of the day. Based on these results, we propose guidelines for visual search for tropical web spiders.