A Robinson 125-W mercury-vapour light-trap was operated at an exposed site near Cranfield aerodrome, south-eastern England. The effects of moon-light, temperature and wind speed on light-trap catch were investigated. The dependent variable, In (catch +1), was compared with the independent variables moonlight illuminance, air temperature and wind speed, using multiple regression analysis. An algorithm was developed and used to estimate moonlight illuminance. Increases in mean illuminance and mean wind speed were associated with a decreased light-trap catch. Increases in mean temperature were associated with an increased catch. When the moth catch was broken down into individual families and species, other trends became apparent. Catches of noctuid moths were unaffected by variation in mean illuminance, although they were influenced by variation in mean temperature and mean wind speed. On the other hand, catches of crambine moths were very sensitive to fluctuations in mean illuminance and mean wind speed although unaffected by changes in mean temperature. The multiple regression equations were used to predict future light-trap catches. The correlations between observed and predicted In (catch +1) for (a) all moths, (b) noctuid moths and (c) crambine moths, were very good. It is suggested that the observed moonlight trend for crambine moths is a consequence of changing behaviour as moonlight illuminance changes rather than the alternative of changing light-trap effectiveness.