Any regular sampling of an insect population results in a series of estimates of the number of each development stage (egg, larval instar, etc.) on successive sampling days. Often many of these stages occur simultaneously owing to a long period of oviposition, hatching, or emergence from hibernation. The analysis of data from this type of population is difficult, since the numbers of any one stage are being reduced by moulting and death, and at the same time are being increased by oviposition, hatching or moulting. Two methods for the estimation of the total number of individuals entering the population and the mortality in each stage from such data, have been described. The first (Richards & Waloff 1954) is based on the calculation of the slope of the fall in the population after the peak in numbers have been reached. By taking logarithms of the numbers, the fall-off will be converted into an approximately straight line, and if this line is produced back to the beginning of the generation, the mortality and the initial number of individuals may be estimated. This method works well for some populations, in which there is a rapid build up and a well-defined peak in numbers. Richards, Waloff & Spradbery (1960) describe a second method by which it is possible to analyse data from a population with a prolonged build-up and an ill-defined peak in numbers. For this, however, the initial number of the population and the duration of each instar must be k_own. A third method is described here which has certain advantages over the other two. In the first place, the duration of each instar and the initial number of individuals do not have to be known. Secondly, this method is independent of the rate of build-up of the population, although this rate must be known. This analysis is applicable to insects with any number of generations a year provided that the generations are distinct, that is, provided that a particular stage in one generation does not overlap the same stage in the next. If it is assumed that the mortality during each larval instar is constant, then the population changes between sampling days may be represented in terms of the accumulated mortalities of the instars present. The number of any one stage dying between sampling days is given by the product of the number of that stage present, the fraction dying per day and the number of days. For simplicity the number present may be taken as the average of the numbers on the