The principles of pattern formation in insects have been studied extensively using classical experimental approaches. In Drosophila, a powerful combination of genetics and transplantation experiments, as well as molecular biology, has helped to elucidate the mechanisms that operate during oogenesis to establish a set of positional cues required for axis determination in the early embryo. These studies suggest the following model: for the anteroposterior axis of the embryo, three groups of maternal genes define three largely independent systems that determine (1) the anterior segmented region of head and thorax, (2) the posterior segmented region of the abdomen, and (3) the terminal non-segmented regions of acron and telson. In contrast, the dorsoventral egg axis appears to require only one system. In each of the four systems, one key gene has an active product that is unequally distributed in the egg. This product provides the spatial signal for the region-specific activation of the transcription of at least one zygotic target gene. The other members within each group serve accessory functions such as determining the correct spatial distribution of the key gene products or controlling their localized activation. The unique expression patterns of the individual zygotic target genes provide a coarse spatial framework which is then refined by the action and interaction of zygotic genes. The notion of three independent systems determining the anteroposterior axis is at variance with a previous model (Meinhardt 1977, Nüsslein-Volhard 1979) of only one gradient, with a high point at the posterior pole, determining a series of states in a concentration-dependent manner. Concentration-dependent determination of more than one quality is likely to occur in the anterior and the dorsoventral system. In contrast, position and polarity within the posterior pattern appear to depend largely on the interaction between gap genes expressed in neighbouring regions rather than on the concentration of the posterior signal.