1. A priori, there are no obvious reasons why patterns should exist in the frequency of density dependence across insect orders. However, orders may reflect related factors which influence population regulation (e.g. life-history patterns and ecology) and are difficult to quantify. The frequency of occurrence of density dependence is compared in 171 time series (of ten or more generations) from Lepidoptera, Hemiptera, Diptera, Odonata, Hymenoptera and Coleoptera. A posteriori attempts are made to identify the cause of observed patterns. 2. Bulmer's (1975) test found non-delayed density dependence more frequently in Odonata than Lepidoptera and Hymenoptera, which in turn showed nondelayed density dependence more frequently than Diptera, Hemiptera and Coleoptera. Similarly, detection was greater for Odonata than other orders using Dennis & Taper's (1993) test for density dependence and Crowley's (1992) test for attraction. Varley & Gradwell's (1960) test found density dependence less frequently in Hemiptera than other orders. These differences were independent of time series length, temporal trends and numbers of generations per year. 3. The reasons for observed patterns in detection of density dependence (and attraction) in insect orders are not clear; however, plausible explanations are differences in: (i) intrinsic growth rate, which is correlated with body size (although evidence to support this hypothesis is weak); (ii) the sampling method used; or (iii) whether individuals come from a single population or many populations. 4. Using Turchin's (1990) test, delayed (lag 2) density dependence was detected most frequently in Hymenoptera, which often show delayed diapause or are parasitoids.
