From 1954 to 1963, Tipula paludosa populations in Northumberland increased every year except 1955 and 1959 when spectacular crashes occurred between August (adult stage) and December (third instar). The years 1955 and 1959 were unusual only in that their rainfalls during August and September were abnormally low. Oviposition begins in the first half of August. It was therefore concluded tentatively that the crashes in 1955 and 1959 were due to excessive mortality from desiccation in the egg and/or first instar stages. A field experiment in 1961, which included simulations of the 1955 and 1959 rainfall conditions, led to the firm conclusion that the crashes of 1955 and 1959 were due mainly to excessive mortality from desiccation in the egg stage. The implications of this work for prediction of population trends are indicated.