Video was used to provide objective measurements of the behaviour of tsetse around an electric net. Data were collected on: (1) the number and behaviour of flies which actively avoided the net; (2) the number and fate of flies which hit the net; (3) the distribution on the ground around the net of electrocuted flies. This allowed estimates to be made of net efficiency (=No. of flies sampled/No. sampled + No. escaped%). The three components of an electric net system (netting/grids, inverter-transformer (‘spark box’) and battery) were varied to determine the effect of each on efficiency: neither the condition (age) of the netting/grids nor the design or condition of the ‘spark boxes’ had a significant effect on efficiency, but efficiency was lower with partly-charged batteries than with fully-charged ones. When account was not taken of flies which behaviourally avoided the net, the best system had an efficiency of 55% (range 49–61%), (cf. a previous estimate of 94%); when the behavioural avoiders were included in the estimate the efficiency was only 46% (range 43–49%). The efficiency of the system is discussed in terms of the way the high voltage across the electric grid is generated, and also how it might affect interpretation of sample data.