The definition and measurement of body condition are central to many ecological and evolutionary studies, yet the fundamental issue of whether variation in condition is inherited genetically remains unresolved. Furthermore, confusion has been caused through diverse uses of the word 'condition' itself. In this paper, we introduce the terms 'broad-sense condition' and 'narrow-sense condition' in an attempt to reduce this confusion. Because of the difficulty of measuring condition, field biologists use the mass relative to body size as a convenient index of condition. Such indices have also been used to estimate the heritability of condition. Using data from a wild great tit Parus major population, we demonstrate that a condition index may stiU contain residual variance that is due to body size. Hence, using a condition index to study the heritability of condition can give misleading results because that residual variance may itself be heritable. T o avoid this problem, direct measures of condition, such as reserve tissues, should be used. We provide the first heritability estimates of direct measures of condition of full-grown birds in a wild population by the regression of mean offspring (both in winter, and when breeding) on mean parent (when breeding) values of two reserve tissues, fat and pectoral muscle, which are independent of body-size. A great tit's fat and muscle reserves in winter were significantly correlated with their respective values when the birds bred. However there was no evidence that the level of fat reserves in winter resembled those of the parents (when breeding), more than expected by chance. Slight resemblances between parents and offspring were detected for muscle in winter and for fat reserves when breeding. Muscle size in the breeding season provided stronger evidence for the inheritance of condition. 0 2000 The Linnean Society of London ADDITIONAL KEY WORDS:-inheritancefatmusclematernal effects -Wytham.