Statistically meaningful comparisons of fossil assemblages can be made by calculating cluster-confidence intervals around mean species abundances using data from replicate bulk samples. However, different levels of precision and reliability are obtained from cluster-confidence intervals depending on how sampling effort is distributed when collecting replicate samples. If the species-abundance distributions at a locality are patchy, then both the precision and the reliability of cluster-confidence intervals are increased by distributing sampling effort among a larger number of bulk samples. Collecting larger numbers of samples does not significantly increase the sampling effort if the total number of individuals collected remains the same. Cluster-confidence intervals (CIs) around mean species abundances at a locality provide a way to assess the magnitude of differences observed between localities and can be used to assess the significance of the absence of rare species. Conclusions derived from cluster-CIs regarding the significance of differences in observed species abundances are similar to those obtained using analysis of variance (ANOVA). Cluster-CIs provide a powerful and easily applied tool for assessing the significance of differences in species abundances and paleo-community composition observed in the fossil record.