The paper describes the layout and analysis of a test of two types of insect light-trap, the Rothamsted and the Robinson, and two types of illumination. The layout consisted of four variations of the sequence of six nights X A B O D D (X being a night of no trapping and A-D the four traps) in each of four sites in a small woodland. At the end of a complete cycle of 24 nights there had been four nights without trapping; each trap had been an equal number of times in each site; and each trap had followed all four trap-types, and also no-trapping, an equal number of times. The complete cycle was repeated four times at intervals between May and September. All insects captured were sorted into orders and counted and each order was analysed separately. By analysis of variance (after transforming the numbers caught to a logarithmic scale) it was possible to show separately the effect of differences, (1) between nights, (2) between sites, (3) between types of trap, (4) between types of light, and, (5) between cycles, and also the effect (if any) of the previous night's trapping. It was found that ultra-violet light was more efficient for all orders than ordinary electric light, that the Robinson type of trap was more efficient for the larger Lepidoptera but much less efficient for the small Diptera, but that there was no regular effect of the previous night's trapping. The largest source of variation was the difference between nights: difference between locations was small.