SUMMARY1. Fluctuations of the numbers of aphids in the air have been regarded in the past as due mainly to weather‐controlled changes in flight behaviour. Migration has been envisaged largely as a calm‐weather phenomenon, whose intensity is limited mainly by the effects of wind‐speed on take‐off behaviour. I have called this the ‘flight‐activity hypothesis’; it developed from both laboratory and field work and has been in vogue for about 20 years.2. This hypothesis does not fully explain changes in numbers of aphids in flight. It is apparently inconsistent with the idea of large‐scale windborne migrations. Many other discrepancies are discussed.3. When studying the changes in numbers of aphids migrating, it is helpful to distinguish between aphids migrating from plants on which they bred and those flying from other places subsequently. Aphids from these two classes together make up the general aphid population in the air. The failure to make this distinction in the past and to appreciate that the factors controlling the numbers flying in the two classes are very different, has led to much confusion.4. A new hypothesis of numbers ofAphis fabaemigrating is developed. But it is at present restricted to those flying from breeding sites; in this respect it is considered to be applicable to many species of aphids. The sequence of events is briefly thus: moulting of the winged nymphs into alatae is usually most intense early in the morning; usually one or more additional moulting peaks occur later in the day. The alatae reach flight maturity usually within 24 hr. after moulting. They then usually fly away in ‘flushes’, on the first migratory flight. The numbers in each ‘flush’ depend on the sizes of previous moulting peaks and on the lengths of both the obligatory period for maturation and on the facultative period which follow; thus the effects of vagaries in flight behaviour tend to be obscured. Therefore, the process is only weakly correlated with current weather factors.The numbers of aphids on subsequent flights will also depend on previous population (moulting) changes as well as on behaviour. Aphids lose their ability to fly owing to autolysis of the flight muscles a few days after they have left the original host, and the rapidity with which this occurs will exert a considerable effect on the numbers of aphids on flights subsequent to the first migration and on their ability to spread virus diseases; the quality of their flight may also be affected.Little is known of the biology of aphids after the first migratory flight, nor of the relative effects of the factors affecting their numbers in the air.5. It is suggested that a considerable proportion of all the aphids in the air are on their first migratory flight; and that the number on subsequent flights may be more limited than seems to have been envisaged.6. The flight‐activity hypothesis is reviewed critically and in detail. Its defects are mainly three‐fold.(a) High local concentrations of aphids occurring in occasional calm periods have been thought to account for most of the aphids migrating. This is not so. Most migration is made up of the lower densities on the more numerous windier occasions.(b) Changes in numbers of aphids in the air from day to day were thought to be due mainly to variation in flight behaviour. They are, in fact, mainly collective‐population changes and subject to different laws than those applying only at the individual level.(c) Errors in measurement and treatment of data.The data of the previous authors are analysed in detail, to illustrate these errors and to disprove some previous contentions.7. Various aspects of aphid dispersal are discussed in the light of old and new hypotheses; these are population change and migration; active and passive flight; migratory and non‐migratory flight; the supposed effects of humidity; so‐called ‘optimum conditions’ for flight. The significance of these aspects to the spread of virus diseases is continually borne in mind.It is a pleasure to acknowledge the help from discussions which I have had with my colleagues Dr J. S. Kennedy, Dr L. Broadbent, Mr L. R. Taylor; also Dr E. Haine and Mr Bruce Johnson who, in addition, have helped respectively with translations and