We assessed the relative magnitude of various factors (year, preservation method, continent, investigator, and taxonomic level) affecting prediction of invertebrate dry mass (DM) in light of the variability of assessments of invertebrate density. We developed 34 length (L)–DM relationships for Oligochaeta and 17 freshwater invertebrate families belonging to Mollusca, Crustacea, and Insecta. Comparison of our predicted DM for reference-size individuals with values from 120 other published equations revealed that 31% of predicted values were within our 95% CI and 73% were within a 2× DM range (i.e., between 0.5× and 2×). Interannual differences in exponent (slope) or scale factor (intercept) of L–DM relationships were detected for 6 of the 7 taxa investigated, but represented only 3% of total variance in predicted DM. Similarly, preservation methods and measured body dimension each accounted for a small (0–3%) fraction of total variance. Variation among investigators (12–50%) and continents (1–17%) were more important and might have reflected methodological or regional and latitudinal differences. Increasingly precise taxonomical levels explained progressively lower proportions of the total variance, a result indicating that family or a more precise taxonomic level provided a robust estimate of most invertebrate DM even if the equations were derived from other sites. However, overall variability induced by L–DM relationships was smaller than variability in total invertebrate density among replicate samples (coefficient of variation [CV]  =  19–97%), a result indicating that more effort should be devoted to improving the accuracy of invertebrate density estimates than to developing site-specific L–DM relationships to assess benthic biomass in freshwater.