The study evaluates various ecological diversity measures for bacterial communities, focusing on species richness, evenness, and abundance models. Bacteria were classified into operational taxonomic units (OTUs) using amplified ribosomal DNA restriction analysis (ARDRA) of 236 clones from contaminated and control soils. The contaminated soil showed reduced diversity, reflected in diversity indices. The number of clones analyzed and the weighting of rare vs. abundant OTUs are critical factors in selecting measures. The log series index (α), Q statistic (with ≥50% coverage), Berger–Parker, and Simpson's indices were preferred, though their ecological relevance may be limited. The Shannon–Wiener and evenness indices were also considered, despite potential inaccuracies. For extrapolation, the log series distribution is best, while non-parametric methods like Chao 1 show promise for estimating total OTU richness. Species abundance models, despite coverage issues, are useful as they address the whole distribution, aiding comparison. The log and log normal models fit the contaminated soil, while none fit the control soil due to single-occurrence OTUs. The Shannon index (H') is widely used but has unclear meaning and is sensitive to rare OTUs. It is positively correlated with diversity and evenness but may be an underestimate due to incomplete coverage. The evenness index (E) is also sensitive to rare OTUs but may overestimate true values with incomplete coverage. The Q statistic focuses on intermediate abundance OTUs but is unreliable with low coverage. Simpson's (D) and Berger–Parker (d) indices are heavily weighted by dominant OTUs and are less affected by coverage. Species abundance models provide a comprehensive view of diversity, avoiding oversimplification. The study highlights the importance of considering coverage and the limitations of indices in assessing bacterial diversity. The results suggest that non-asymptotic models and non-parametric estimators like Chao 1 are more reliable for diverse communities. The study concludes that diversity indices and models should be selected based on their suitability for the specific ecological context and data characteristics.The study evaluates various ecological diversity measures for bacterial communities, focusing on species richness, evenness, and abundance models. Bacteria were classified into operational taxonomic units (OTUs) using amplified ribosomal DNA restriction analysis (ARDRA) of 236 clones from contaminated and control soils. The contaminated soil showed reduced diversity, reflected in diversity indices. The number of clones analyzed and the weighting of rare vs. abundant OTUs are critical factors in selecting measures. The log series index (α), Q statistic (with ≥50% coverage), Berger–Parker, and Simpson's indices were preferred, though their ecological relevance may be limited. The Shannon–Wiener and evenness indices were also considered, despite potential inaccuracies. For extrapolation, the log series distribution is best, while non-parametric methods like Chao 1 show promise for estimating total OTU richness. Species abundance models, despite coverage issues, are useful as they address the whole distribution, aiding comparison. The log and log normal models fit the contaminated soil, while none fit the control soil due to single-occurrence OTUs. The Shannon index (H') is widely used but has unclear meaning and is sensitive to rare OTUs. It is positively correlated with diversity and evenness but may be an underestimate due to incomplete coverage. The evenness index (E) is also sensitive to rare OTUs but may overestimate true values with incomplete coverage. The Q statistic focuses on intermediate abundance OTUs but is unreliable with low coverage. Simpson's (D) and Berger–Parker (d) indices are heavily weighted by dominant OTUs and are less affected by coverage. Species abundance models provide a comprehensive view of diversity, avoiding oversimplification. The study highlights the importance of considering coverage and the limitations of indices in assessing bacterial diversity. The results suggest that non-asymptotic models and non-parametric estimators like Chao 1 are more reliable for diverse communities. The study concludes that diversity indices and models should be selected based on their suitability for the specific ecological context and data characteristics.