Abstract. 1. Farrow light traps have been used in Australia since the 1970s to monitor locusts only, but catch variability has not been explained. Four light traps were operated at North Bourke, northern New South Wales, to study catch variability.
2. Habitat did not influence catch composition (non-metric multidimensional scaling/analyses of similarity) or similarity. Assemblages of Orthoptera differed only modestly between December and March, whereas assemblages of macrolepidoptera differed strongly between December, March and April.
3. Twenty-two species of Orthoptera were caught and seven analysed. Kruskal–Wallis ranking tests showed that the largest catches of a species of cricket were obtained when the dominant wind direction was from the SE, whereas the largest catches of three acridid species were obtained when the wind was from the NW. Catch variability of two acridids could be explained by temperature (Generalised linear models); the smallest acridid by wind speed and of the cricket and two acridids by moon surface illumination (K).
4. Eighty-two species of macrolepidoptera were identified and 17 analysed. The largest catches of a noctuid were obtained when the wind was from the SW; an oenosandrid was only caught when the wind was from the SE or NE. Catch variability of two noctuids, an arctiid and an oenosandrid could be explained by temperature; three noctuids, an arctiid, a cossid and a psychid by relative humidity; a noctuid by wind speed; and two noctuids and two oenosandrids by K.
5. To facilitate comparisons of catches and interpret long-term population trends, catch variability should be standardised for meteorological/lunar conditions.