Energy budgets are based upon the equation C=P P + R + FU, where C = Consumption, P = Production, R = Respiration, and FU= Rejecta (comprising F, egesta, and U, excreta). In studies of insect energetics where all four quantities have been measured, the completed budget rarely balances. Wightman (1981) reviewed energy budgets of insects, spiders, and marine invertebrates and evaluated errors that arise in estimation of each of the components. By process of elimination, he traced the imbalance to inaccuracy in the R term which, when measured with a respirometer (Rm) is almost invariably less than when calculated as CPFU (=R,). Wightman (1977, 1981) associated differences between Rm and RC with the 'unnatural' conditions associated with the respirometer, but he preferred to leave these 'flask effects' undefined. In this paper I (1) reexamine the arguments by which Wightman (1981) excluded errors in C, FU, and P to isolate inaccuracy in R as the cause of budgetary imbalance, and (2) propose an explanation for discrepancies in Rm and Rc. What are the possible biases in estimation of C, FU, P, and R? Consumption estimates may be biased due to failure to correct for changes in the uneaten food during the experimental period. Weight loss due to respiration of the diet in the gravimetric method (Axelsson and Agren 1979) or leaf shrinkage in the planimetric method (Axelsson et al. 1974, Axelsson et al. 1975) may lead to overestimation of C. Weight gain in leaves due to photosynthesis may lead to underestimation of C (Edwards and Wightman 1984). The term rejecta includes both undigested food and metabolic products, such as urine, digestive enzymes, and peritrophic membrane. The contribution of excretory products to the weight of the rejecta is generally negligible (Waldbauer 1968), so egestion may be estimated from the weight of the rejecta without serious exaggeration. Egestion may be underestimated due to evaporation of components of moist feces, activity of decomposers, or coprophagy by the feeding insect (Wightman 1981). Production may be underestimated due to failure to include exuviae (Carefoot 1977) or weight lost prior to the molt (Witkowski and Kosior 1974), or overestimated due to inclusion of food in the guts of feeding stages (Wightman 1981, Edwards and Wightman 1984). Respiration measured in a respirometer consistently underestimates respiration calculated from laboratory and field feeding studies (Wightman 1981). Closed system respirometers commonly used for such measurements have the disadvantages that animals respire carbon dioxide-free air, often have their locomotory activity restricted, or have a declining rate of oxygen consumption due to depletion of the oxygen in the chamber or starvation of the animal (Lawton and Richards 1970, Wightman 1977). Restriction of activity, depletion of available oxygen, and starvation would lead to low estimates of respiration, while breathing CO2-free air may cause insect respiration to fluctuate erratically (Wightman 1977). Which of these biases are found in studies yielding complete energy budgets? Aided by a recent review of insect energetics (Wiegert and Petersen 1983), I have amended and extended the list of complete energy budgets for invertebrates compiled by Wightman (1981) (see Table 1). I have indicated the extent of budgetary imbalance and the presence and direction of bias in estimation of C, FU, P, and Rm. Budgetary balance may be symbolized by the ratio of RcRm, although I do not intend to imply that errors in the budget reside solely in the R terms. The ratio ranges from 6.02 (indicating wide discrepancy) to a value of 1.00 (indicating equivalence of the two estimates of R). Improved balance is associated with two conditions (1) low respiration of the diet or inclusion of appropriate controls prevents overestimation of consumption (and hence reduces estimates of Re), (2) measurement of insect respiratory rate during or shortly after feeding elevates the value of Rm over that at rest, bringing it more in line with R,. Grouping the studies based on whether they contain, biases in both, one, or neither of the budget components C and Rm, we find (1) there are significant differences in R,/Rm among groups (Kruskal-Wallis test, H= 13.72, p< 0.005) and (2) the mean value of Rc/Rm declines 3.20, 1.46, 1.24 respectively for groups containing both, one, or neither of these errors. Is correction of biases in estimation of C and Rm sufficient to balance the energy budgets reviewed in Table 1 ? First, consider consumption. If the diet loses weight during the feeding trial due to its own respiration, then a simple difference in the amount of food at the beginning and end of the interval will overestimate animal consumption. Axelsson and Agren (1979) give a formula for Q, a correction factor that applied to weight loss of plant tissue can account for plant respiration when calculating consumption. Ex-