Accurate estimates of invertebrate biomass are essential for quantifying community structure, food web dynamics and energy flow in terrestrial ecosystems. In this paper, length-mass and length × width-mass regressions were carried out for 18 invertebrate taxonomic groups collected from the canopy of an Australian tropical rainforest. In an additional analysis, invertebrates were divided among seven body shape categories based on the ratio of body length to body width (from short and squat to long and thin) in an attempt to develop accurate equations for estimating biomass that can be applied to any taxonomic group in any locality. For most groups, the inclusion of body width to the predictor variable improved the model, confirming that body shape is an important factor in the accuracy of biomass estimations. The most accurate method for estimating invertebrate biomass was the use of taxon-specific equations, followed by equations based on body shape. Single whole-fauna equations were very inaccurate for estimating biomass, especially for insects that are either very squat or very long and thin. In accordance with previous studies, it was concluded that the most accurate method for estimating invertebrate biomass from proxy body measurements is the use of taxon-specific regression equations, especially those that incorporate body width in the model. However, equations based on body shape categories may be useful for estimating the biomass of groups for which no length-mass relationship has been determined, while single, whole-fauna equations should be avoided.